IJACSA Volume2No10

(IJACSA) International Journal of Advanced Computer Science and Applications,
Vol. 2, No. 10, 2011

(i)
www.ijacsa.thesai.org
IJACSA Editorial
From the Desk of Managing Editor
It is a pleasure to present our readers with the October 2011 Issue of International Journal of Advanced Computer Science and
Applications (IJACSA).
The renaissance stimulated by the field of Computer Science is generating multiple formats and channels of communication and
creativity. IJACSA is one of the most prominent publications in the field and engaging the ubiquitous spread of subject
knowledge with effectiveness in all classes of audience. Nevertheless, the promise of increased engagement requires that we
consider how this might be accomplished, delivering up-to-date and authoritative coverage of advanced computer science and
applications.
The journal has a wide scope ranging from the many facets of methodological foundations to the details of technical issues and
the aspects of industrial practice. It includes articles related to research findings, technical evaluations, and reviews. In addition
it provides a forum for the exchange of information on all aspects.
The editorial board of the IJACSA consists of individuals who are committed to the search for high-quality research suitable for
publication. These individuals, working with the editor to achieve IJACSA objectives, assess the quality, relevance, and
readability of individual articles.
The contents include original research and innovative applications from all parts of the world. This interdisciplinary journal has
brought together researchers from academia and industry as well as practitioners to share ideas, problems and solutions
relating to computer science and application with its convergence strategies, and to disseminate the most innovative research.
As a consequence only 28% of the received articles have been finally accepted for publication.
Therefore, IJACSA in general, could serve as a reliable resource for everybody loosely or tightly attached to this field of science.
The published papers are expected to present results of significant value to solve the various problems with application services
and other problems which are within the scope of IJACSA. In addition, we expect they will trigger further related research and
technological improvements relevant to our future lives.
We hope to continue exploring the always diverse and often astonishing fields in Advanced Computer Science and Applications.
Thank You for Sharing Wisdom!

Managing Editor
IJACSA
Volume 2 Issue 10, October 2011
[email protected]
ISSN 2156-5570 (Online)
ISSN 2158-107X (Print)
2011 The Science and Information (SAI) Organization
Vol. 2, No. 10, 2011
(ii)
IJACSA Associate Editors
Dr. Zuqing Zhu
Service Provider Technology Group of Cisco Systems, San Jose
Domain of Research: Research and development of wideband access routers for hybrid fibre-
coaxial (HFC) cable networks and passive optical networks (PON)
Dr. Ka Lok Man
Department of Computer Science and Software Engineering at the Xi'an Jiaotong-Liverpool
University, China
Domain of Research: Design, analysis and tools for integrated circuits and systems; formal
methods; process algebras; real-time, hybrid systems and physical cyber systems;
communication and wireless sensor networks.
Dr. Sasan Adibi
Technical Staff Member of Advanced Research, Research In Motion (RIM), Canada
Domain of Research: Security of wireless systems, Quality of Service (QoS), Ad-Hoc Networks,
e-Health and m-Health (Mobile Health)
Dr. Sikha Bagui
Associate Professor in the Department of Computer Science at the University of West Florida,
Domain of Research: Database and Data Mining.
Dr. T. V. Prasad
Dean, Lingaya's University, India
Domain of Research: Bioinformatics, Natural Language Processing, Image Processing, Expert
Systems, Robotics
Dr. Bremananth R
Research Fellow, Nanyang Technological University, Singapore
Domain of Research: Acoustic Holography, Pattern Recognition, Computer Vision, Image
Processing, Biometrics, Multimedia and Soft Computing
Vol. 2, No. 10, 2011
(iii)
IJACSA Reviewer Board

A Kathirvel
Karpaga Vinayaka College of Engineering and
Technology, India
Abbas Karimi
I.A.U_Arak Branch (Faculty Member) & Universiti
Putra Malaysia
Dr. Abdul Wahid
Gautam Buddha University, India
Abdul Khader Jilani Saudagar
Al-Imam Muhammad Ibn Saud Islamic University
Abdur Rashid Khan
Gomal Unversity
Dr. Ahmed Nabih Zaki Rashed
Menoufia University, Egypt
Ahmed Sabah AL-Jumaili
Ahlia University
Md. Akbar Hossain
Aalborg University, Denmark and AIT, Greeceas
Albert Alexander
Kongu Engineering College,India
Prof. Alcinia Zita Sampaio
Technical University of Lisbon
Amit Verma
Rayat & Bahra Engineering College, India
Ammar Mohammed Ammar
Department of Computer Science, University of
Koblenz-Landau
Arash Habibi Lashakri
University Technology Malaysia (UTM), Malaysia
Asoke Nath
St. Xaviers College, India
B R SARATH KUMAR
Lenora College of Engineering, India
Binod Kumar
Lakshmi Narayan College of Technology, India
Bremananth Ramachandran
School of EEE, Nanyang Technological University
Dr.C.Suresh Gnana Dhas
Park College of Engineering and Technology, India
Mr. Chakresh kumar
Manav Rachna International University, India
Chandra Mouli P.V.S.S.R
VIT University, India
Chandrashekhar Meshram
Shri Shankaracharya Engineering College, India
Constantin POPESCU
Department of Mathematics and Computer
Science, University of Oradea
Prof. D. S. R. Murthy
SNIST, India.
Deepak Garg
Thapar University.
Prof. Dhananjay R.Kalbande
Sardar Patel Institute of Technology, India
Dhirendra Mishra
SVKM's NMIMS University, India
Divya Prakash Shrivastava
EL JABAL AL GARBI UNIVERSITY, ZAWIA
Dragana Becejski-Vujaklija
University of Belgrade, Faculty of organizational
sciences
Fokrul Alom Mazarbhuiya
King Khalid University
G. Sreedhar
Rashtriya Sanskrit University
Ghalem Belalem
University of Oran (Es Senia)
Hanumanthappa.J
University of Mangalore, India
Dr. Himanshu Aggarwal
Punjabi University, India
Huda K. AL-Jobori
Ahlia University
Dr. Jamaiah Haji Yahaya
Northern University of Malaysia (UUM), Malaysia
Jasvir Singh
Communication Signal Processing Research Lab
Jatinderkumar R. Saini
S.P.College of Engineering, Gujarat
Prof. Joe-Sam Chou
Nanhua University, Taiwan
Dr. Juan Jos Martnez Castillo
Yacambu University, Venezuela
Dr. Jui-Pin Yang
Vol. 2, No. 10, 2011
(iv)
Shih Chien University, Taiwan
Dr. K.PRASADH
Mets School of Engineering, India
Ka Lok Man
Xian Jiaotong-Liverpool University (XJTLU)
Dr. Kamal Shah
St. Francis Institute of Technology, India
Kodge B. G.
S. V. College, India
Kohei Arai
Saga University
Kunal Patel
Ingenuity Systems, USA
Lai Khin Wee
Technischen Universitt Ilmenau, Germany
Latha Parthiban
SSN College of Engineering, Kalavakkam
Mr. Lijian Sun
Chinese Academy of Surveying and Mapping, China
Long Chen
Qualcomm Incorporated
M.V.Raghavendra
Swathi Institute of Technology & Sciences, India.
Madjid Khalilian
Islamic Azad University
Mahesh Chandra
B.I.T, India
Mahmoud M. A. Abd Ellatif
Mansoura University
Manpreet Singh Manna
SLIET University, Govt. of India
Marcellin Julius NKENLIFACK
University of Dschang
Md. Masud Rana
Khunla University of Engineering & Technology,
Bangladesh
Md. Zia Ur Rahman
Narasaraopeta Engg. College, Narasaraopeta
Messaouda AZZOUZI
Ziane AChour University of Djelfa
Dr. Michael Watts
University of Adelaide, Australia
Miroslav Baca
University of Zagreb, Faculty of organization and
informatics / Center for biomet
Mohamed Ali Mahjoub
Preparatory Institute of Engineer of Monastir
Mohammad Talib
University of Botswana, Gaborone
Mohammed Ali Hussain
Sri Sai Madhavi Institute of Science & Technology
Mohd Helmy Abd Wahab
Universiti Tun Hussein Onn Malaysia
Mohd Nazri Ismail
University of Kuala Lumpur (UniKL)
Mueen Uddin
Universiti Teknologi Malaysia UTM
Dr. Murugesan N
Government Arts College (Autonomous), India
Nitin S. Choubey
Mukesh Patel School of Technology Management &
Eng
Dr. Nitin Surajkishor
NMIMS, India
Paresh V Virparia
Sardar Patel University
Dr. Poonam Garg
Institute of Management Technology, Ghaziabad
Raj Gaurang Tiwari
AZAD Institute of Engineering and Technology
Rajesh Kumar
National University of Singapore
Rajesh K Shukla
Sagar Institute of Research & Technology-
Excellence, India
Dr. Rajiv Dharaskar
GH Raisoni College of Engineering, India
Prof. Rakesh. L
Vijetha Institute of Technology, India
Prof. Rashid Sheikh
Acropolis Institute of Technology and Research,
India
Ravi Prakash
University of Mumbai
Rongrong Ji
Columbia University
Dr. Ruchika Malhotra
Delhi Technological University, India
Dr.Sagarmay Deb
University Lecturer, Central Queensland University,
Australia
Vol. 2, No. 10, 2011
(v)
Saleh Ali K. AlOmari
Universiti Sains Malaysia
Dr. Sana'a Wafa Al-Sayegh
University College of Applied Sciences UCAS-
Palestine
Santosh Kumar
Graphic Era University, India
Sasan Adibi
Research In Motion (RIM)
Saurabh Pal
VBS Purvanchal University, Jaunpur
Seyed Hamidreza Mohades Kasaei
University of Isfahan
Shahanawaj Ahamad
The University of Al-Kharj
Shaidah Jusoh
University of West Florida
Sikha Bagui
Zarqa University
Dr. Smita Rajpal
ITM University
Suhas J Manangi
Microsoft
SUKUMAR SENTHILKUMAR
Universiti Sains Malaysia
Sunil Taneja
Smt. Aruna Asaf Ali Government Post Graduate
College, India
Dr. Suresh Sankaranarayanan
University of West Indies, Kingston, Jamaica
T C.Manjunath
Visvesvaraya Tech. University
T V Narayana Rao
Hyderabad Institute of Technology and
Management, India
T. V. Prasad
Lingaya's University
Taiwo Ayodele
Lingaya's University
Totok R. Biyanto
Infonetmedia/University of Portsmouth
Varun Kumar
Institute of Technology and Management, India
Vellanki Uma Kanta Sastry
Sreeneedhi
Dr. V. U. K. Sastry
SreeNidhi Institute of Science and Technology
(SNIST), Hyderabad, India.
Vinayak Bairagi
Sinhgad Academy of engineering, India
Vitus S.W. Lam
The University of Hong Kong
Vuda Sreenivasarao
St.Marys college of Engineering & Technology,
Hyderabad, India
Y Srinivas
GITAM University
Mr.Zhao Zhang
City University of Hong Kong, Kowloon, Hong Kong
Zhixin Chen
ILX Lightwave Corporation
Zuqing Zhu
University of Science and Technology of China
Vol. 2, No. 10, 2011
(vi)
CONTENTS

Paper 1: Data Mining for Engineering Schools
Authors: Chady El Moucary
PAGE 1 9

Paper 2: Model of Temperature Dependence Shape of Ytterbium -doped Fiber Amplifier Operating at
915 nm Pumping Configuration
Authors: Abdel Hakeim M. Husein, Fady I. EL-Nahal

PAGE 10 13

Paper 3: Healthcare Providers Perceptions towards Health Information Applications at King Abdul-Aziz
Medical City, Saudi Arabia
Authors: Abeer Al-Harbi
PAGE 14 22

Paper 4: Route Maintenance Approach For Link Breakage Prediction In Mobile Ad Hoc Networks
Authors: Khalid Zahedi, Abdul Samad Ismail
PAGE 23 30

Paper 5: Web Service Architecture for a Meta Search Engine
Authors: K.Srinivas, P.V.S. Srinivas, A.Govardhan
PAGE 31 36

Paper 6: A Statistical Approach For Latin Handwritten Digit Recognition
Authors: Ihab Zaqout
PAGE 37 40

Paper 7: Plant Leaf Recognition using Shape based Features and Neural Network classifiers
Authors: Jyotismita Chaki, Ranjan Parekh
PAGE 41 47

Paper 8: Integrated Information System for reserving rooms in Hotels
Authors: Dr. Safarini Osama
PAGE 48 52

Paper 9: Automatic Classification and Segmentation of Brain Tumor in CT Images using Optimal
Dominant Gray level Run length Texture Features
Authors: A.Padma, R.Sukanesh
PAGE 53 59

Vol. 2, No. 10, 2011
(vii)
Paper 10: An Ontology- and Constraint-based Approach for Dynamic Personalized Planning in Renal
Disease Management
Authors: Normadiah Mahiddin, Yu-N Cheah, Fazilah Haron
PAGE 60 69

Paper 11: Asynchronous Checkpointing And Optimistic Message Logging For Mobile Ad Hoc Networks
Authors: Ruchi Tuli, Parveen Kumar
PAGE 70 76

Paper 12: An Experimental Improvement Analysis of Loss Tolerant TCP (LT-TCP) For Wireless Network
Authors: Md. Abdullah Al Mamun, Momotaz Begum, Sumaya kazary, Md. Rubel
PAGE 77 80

Paper 13: Comparison of Workflow Scheduling Algorithms in Cloud Computing
Authors: Navjot Kaur, Taranjit Singh Aulakh, Rajbir Singh Cheema
PAGE 81 86

Paper 14: Study Of Indian Banks Websites For Cyber Crime Safety Mechansim
Authors: Susheel Chandra Bhatt, Durgesh Pant
PAGE 87 90

Paper 15: An Empirical Study of the Applications of Web Mining Techniques in Health Care
Authors: Dr. Varun Kumar, MD. Ezaz Ahmed
PAGE 91 94

Paper 16: Quality EContent Design using Reusability approach
Authors: Senthil Kumar.J, Dr S.K.Srivatsa
PAGE 95 97

Paper 17: Retrieval of Images Using DCT and DCT Wavelet Over Image Blocks
Authors: H. B. kekre, Kavita Sonawane
PAGE 98 106

Paper 18: Cryptanalysis of An Advanced Authentication Scheme
Authors: Sattar J Aboud, Abid T. Al Ajeeli
PAGE 107 111

Paper 19: Conceptual Level Design of Semi-structured Database System: Graph-semantic Based
Approach
Authors: Anirban Sarkar
PAGE 112 121

Vol. 2, No. 10, 2011
1 | P a g e
Data Mining for Engineering Schools
Predicting Students Performance and Enrollment in Masters Programs

Chady El Moucary
Department of Electrical, Computer and Communication Engineering, Faculty of Engineering
Notre Dame University Louaize (NDU)
North Lebanon Campus P.O. Box 87, Tripoli Municipality Street, Barsa El Koura, Lebanon

Abstract the supervision of the academic performance of
engineering students is vital during an early stage of their
curricula. Indeed, their grades in specific core/major courses as
well as their cumulative General Point Average (GPA) are
decisive when pertaining to their ability/condition to pursue
Masters studies or graduate from a five-year Bachelor-of-
Engineering program. Furthermore, these compelling strict
requirements not only significantly affect the attrition rates in
engineering studies (on top of probation and suspension) but also
decide of grant management, developing courseware, and
scheduling of programs. In this paper, we present a study that
has a twofold objective. First, it attempts at correlating the
aforementioned issues with the engineering students
performance in some key courses taken at early stages of their
curricula, then, a predictive model is presented and refined in
order to endow advisors and administrators with a powerful
decision-making tool when tackling such highly important issues.
Matlab Neural Networks Pattern Recognition tool as well as
Classification and Regression Trees (CART) are fully deployed
with important cross validation and testing. Simulation and
prediction results demonstrated a high level of accuracy and
offered efficient analysis and information pertinent to the
management of engineering schools and programs in the frame of
the aforementioned perspective.
Keywords-component; Educational Data Mining; Classification
and Regression Trees (CART); Relieff tool; Neural Networks;
Prediction; Engineering Students Performance; Engineering
Students Enrollment in Masters Studies.
I. INTRODUCTION
Data mining has attracted exceptionally diversified
businesses for both the descriptive and predictive capabilities it
promises, one of which is Education in its broad fields and
organizational hierarchies [15] [36] [50]. In fact, Education,
nowadays, not only involves the ancestral information- and/or
knowledge- communication and transfer, but has also become a
standalone and comprehensive business with excessive
demands in information handling and analysis, as well as the
management of a deeply spread tree of positions and
interrelated functions [49]. Indeed, Educations features and
attributes have dramatically shifted and augmented to a point
where the integration of technology became inevitable in the
attempt of sustaining a good position and thriving in a highly
competitive and merciless market. This technology not only
involves new teaching methodologies but also osculates with
every single aspect of management of such institutions.
Furthermore, management of large amount of data has
become undeniably forbearing to even expert staff; it even
requires more powerful computational and specifications
requirements when referring to machines and/or algorithms.
Diversified challenges face Education and which fortunately
attracted researchers from different fields of expertise who
keep straining in order to achieve innovative but also intelligent
techniques to help keep up with the pressure and find astute
and reasonable answers to multifaceted questions.
Globalization, International Accreditation, and e-learning have
only added more threads to the pile.
Data mining, which is the science of digging into databases
for information and knowledge retrieval, has recently
developed new axes of applications and engendered an
emerging discipline, called Educational Data Mining or EDM.
This discipline seems to be a lot promising. EDM carries out
tasks such as prediction (classification, regression, and density
estimation), clustering, relationship mining (association,
correlation, sequential mining, and causal data mining),
distillation of data for human judgment, and discovery with
models [1]. Moreover, by exercising EDM, educators and
administrators can tackle both traditional and ad hoc
educational issues and benefit from a good decision-making
tool when facing challenges and/or exploring new horizons in
their specialties. The list is long and requires wide and long
tables to fit in. Nevertheless, in a non-exhaustive list, we can
enumerate the most frequently inquiring subject matters such as
predicting students performance, developing courseware,
students behavioral modeling, strategic planning and
scheduling of programs, supervising attrition rates, and grant
management, etc. In other words, EDM aims at enhancing the
understanding and supervision of learners, teachers, and
administrators domain representation, pedagogical
engagement and behaviors [5] [18] [32] [37] [38] [39].
Remarkable amount of EDM endeavors have been
conducted and published in many journals and conference
proceedings related to, but not limited to, Artificial
Intelligence, Learning Systems, Education, and others. In July
2011 the International Educational Data Mining Society [2]
was founded by the International Working Group on
Educational Data Mining with the main objective of capturing
contributions from the EDM community and offering a forum
for practitioners to impart their labor and exchange their
competencies. It has so far organized four international
conferences where recognized work can be archived in the
Vol. 2, No. 10, 2011
2 | P a g e
Journal of Educational Data Mining or JEDM (ISSN 2157-
2100) [28].
One can find many definitions for data mining in books,
journal papers, and e-articles [11] [12] [13] [14]. They all refer
to data mining as a young and interdisciplinary field in
computer science which is described as an interactive and
iterative process aiming at sundering out/revealing
hidden/unobvious, but existing, patterns, trends and/or
relationships amidst data using statistical and mathematical
procedures with a prime objective of providing decision
support systems with information and knowledge.
Furthermore, data mining is being recently interchangeably
used with what is referred to as Knowledge Discovery in
Database or KDD namely when excessively large data
repositories is being involved. Fig. 1 shows Data Mining
exercise as a step towards KDD [10].
A typical example of inexorable flood of data is the
Europe's Very Long Baseline Interferometry (VLBI), which
has 16 telescopes, each of which produces one Gigabit/second
of astronomical data over a 25-day observation session [6]. An
interesting overview of the largest databases in the world can
be found in [7] [8]. The top-ten lists include The Library of
Congress (LC) with over 130 million items, 530 miles of
shelves, 5 million digital documents, and 20 terabytes of text
data. The Central Intelligence Agency (CIA) possesses
comprehensive statistics on more than 250 countries and
entities (unknown number of classified information). Amazon,
the worlds biggest retail store, maintains over 59 million
active customers ending up with over 42 terabytes of data.
YouTube, the largest video library, observes more than 65,000
videos added each day and encompasses at least 45 terabytes of
videos. ChoicePoint, the business of acquiring information
about the American population (addresses, phone numbers,
driving records, etc.) possesses a database that extends to the
moon and back 77 times and holds over 250 terabytes of
personal data. Sprint, one of the worlds largest
telecommunication companies, offers its services to more than
53 million subscribers with a 2.85 trillion database rows and
70,000 call detail-record insertions per second. Google, the
famous search engine and industry, is subjected to 91 million
searches per day (accounting to 50% of all internet search
activity) and holds more than 33 trillion database entries, etc.
It is spectacularly evident that traditional warehousing
techniques and querying algorithms cannot cope with such
colossal amount of data, thus, new techniques for information
retrieval urged tons of research papers in the data mining/KDD
field. Many strains have been deployed to manipulate and
handle considerable amount of data [20] [35] [47] [48], but in
many applications, the data available only covers parts of the
inference chain from evidence to actions. However, a versed
miner can extrapolate a small amount of initial knowledge into
more knowledge using proficient mining.
In this paper, we will present a study that aims at offering a
reliable and predictive tool for academicians and administrators
working in engineering schools and universities to monitor
students performance at an early stage of their educational
path. The goal is to link this observation/data with students
chances to either finish (succeed) a five-year Bachelor-of-
Engineering program (BE) or enroll in Masters program in a
BS/MS track.
In the section to come, Data Mining will be reviewed and
presented from different perspectives with emphasis on its
various categories, tasks and implementations. In Section III,
we will elaborate on the tool developed and underline data
preparation and attributes selection. Furthermore, the use of
Neural Networks and Classification and Regression Trees
(CART) will be explained and applied with cross-validation
and pruning [40]. Error Histogram and ROC curves will also
be studied in section III. Finally, section IV will portray a
thorough analysis and discussions of the results. The core
objective of the paper will be summarized and a conclusion is
presented in this section as well.
II. DATA MINING
A. A Multifaceted Discipline
Data mining is a twofold discipline in the sense that it
subtends two high-level primary objectives: prediction and
description [10]. Prediction involves using some variables or
fields in the database to predict unknown or future values of
other variables of interest. Description focuses on finding
human-interpretable patterns describing the data. The relative
importance of prediction and description for particular data
mining applications can vary considerably. While in the
context of KDD description tends to be more important than
prediction, prediction is often the primary goal in pattern
recognition and machine learning applications.
B. Learning Approaches and Techniques
Data mining can be described as either supervised or
unsupervised [9] [48] as shown in Fig. 2. Unsupervised data
mining is rather a bottom-up approach that makes no prior
assumptions and aims at discovering relationships in the data.
In this sense, data are allowed to speak for themselves; there is
no distinction between attributes and targets. In this context,
unsupervised data mining is a descriptive approach. Typical
methods and applications are clustering, density estimation,
data segmentation, smoothing, etc. Supervised data mining,
also called direct data mining, aims at explaining those

Figure 1 - Data Mining, a Mandatory Step towards KDD
Vol. 2, No. 10, 2011
3 | P a g e
relationships once they are found. It is rather a predictive
approach where the variables involved are classified as
explanatory and dependent ones, and where the main goal is to
achieve a liaison between them as in Regression Analysis.
Typically, the target variable has to be well specified in
advance and three important steps follow before achieving data
mining or KDD purposes, as illustrated in Fig. 3.
Specifically, data is to be subdivided for training, testing,
and validation. The objective of the training step is to
construct a provisional model that attempts to subtend and/or
engender the hidden relationship between the attributes and the
target variable. The validation step plays a key role in reducing
the overfitting traits of the model in the sense that it helps
reduce the amount of unsatisfactory results or avoid patterns
that are not present in the general dataset (sometimes called flat
file). This would occur if the model has excessive number of
attributes relative to the amount of data collected and available;
the model will exaggerate minor fluctuations in the data and
thus, have poor predictive capacity.
Another typical hitch could be an acquired memory
characteristic that downsizes the model to specific cases in the
training phase. For instance, assume that in the training data all
students who have passed pre-calculus course have succeeded
their graduate studies; we do not want the model to remember
this liaison and create the pattern if the student succeeds the
pre-calculus course, then he/she will succeed the graduate
studies. Instead, the model should apply all patterns found in
the training phase to the future data and thus, acquire a
generalization characteristic/capability. A number of statistical
techniques can be applied to assess the model such as the ROC
(Receiver/Relative Operating Characteristic) curves, which
depict the true positive rate vs. the false positive one as the
threshold of discrimination is modified in the case of a binary
classifier. Finally, testing is used to evaluate, revisit and/or
retrain the constructed model by using new untrained data that
have been held out from the original complete dataset. At this
point, the data miner should be able to ascertain whether the
learned patterns meet the desired standards. If the outcome is
satisfactory, the data miner shall transform the model into
information and knowledge, otherwise, re-training, changing
the pre-processing and/or the data-mining algorithm is
inevitable. This is where the expertise of the data miner plays a
decisive role. In fact, not only this expertise is crucial in
rectifying the path of the mining process, but also in converting
the information into knowledge in the sense of filling the gap
between the past (existing data) and the future (prediction) and
drawing an efficient roadmap (decision making and strategic
planning) using the outcome of the KDD process.
C. Data Mining Tasks
Data mining objectives can be carried out by means of
various procedures, frequently called tasks. Thus a further
categorization of data mining is obtained:
Classification: Typical supervised-learning task where
information is arranged into predefined classes
according to some learnt rules. The learning process
aims at developing a model for predicting/assigning the
class of a new instance. In other words, it is the
generalized application of a known structure to a new
data for mapping and classification purposes. The
most widely used classifiers are Decision Trees,
Bayesian and Neural Networks, etc. This task is
applied to diversified fields of expertise such as,
Speech and Handwriting Recognition, Web Search
Engines, Geostatistics (remote sensing), etc. [42] [44]
Regression: The goal of this task is to achieve a
function (regression function) of the independent
variables that allows computing the conditional
expectation of a dependent variable for prediction and
forecasting exercises based on the minimization of a
certain type of error via an iterative procedure.
Practically, Classification and Regression Trees
(CART) summarize these tasks; classification is
referred to when the target is nominal, whereas
regression is used for continuous values of the target
(infinite number of values).
Clustering: It is a descriptive and typical unsupervised
machine-learning task common to statistical data
analysis wherein a finite set of groups and clusters are
identified to describe the data. It is referred to in
applications such as Image Analysis, Machine
Learning, Biology and Medicine, Pattern Recognition,
Education, Crime Analysis, etc. The most reputed
approaches related to this task are the K-Means [26]
and Fuzzy Clustering techniques [25] [33].
Summarization: Various methods are formulated to
describe the set of data and information in a more

Figure 2 Data Mining Learning Techniques

Figure 3 - Roadmap for an Efficient Predictive Tool

Vol. 2, No. 10, 2011
4 | P a g e
compact representation. It also includes report
generation.
Association Rule Learning or Dependency Modeling:
This task aims at identifying frequent itemsets in a
database and deriving association rules. A local model
is identified and which describes the important
dependencies between variables and datasets. It has
been mainly used in applications involving decisions
about marketing activities in supermarkets. It is also
applied in the fields of Web Mining, Bioinformatics,
etc. The most popular and famous algorithm used in
this field is the Apriori [23] [24] [30] [31].
Outlier/Deviation Detection: Important and significant
deviations in the dataset are reported. It finds
application in Fraud Detection, Intrusion Detection
(Data Security), etc. It is used to increase accuracy by
removing anomalous data from the dataset
(supervised). Popular algorithms are based on K-
Nearest Neighbor, Support Vector Machines, etc. [21]
[22] [34].
Link Analysis: Find relationships amongst richly
structured databases where hidden patterns are
somewhat difficult to be discerned using traditional
statistical approaches [20].
Finally, it is noteworthy to mention that estimation and
prediction are sometimes interchangeably used when dealing
with classification and regression problems. The reality is that
estimation is used when a continuous value is to be forecasted
while prediction is referred to when new data is classified into
one of predefined classes, which are predetermined when
building the model.
D. Data Miner Role
It should be noted that efficient and plausible mining
exercises remain highly dependent from data preprocessing
such as gathering, cleaning, representation, etc. Indeed,
although data mining tools and tasks can be very appealing,
domain-specific skills are required as a prerequisite before
embarking on the trip towards KDD. In other words, human
interface plays a decisive role in somewhat deploying or
transforming data from an opaque entity into a transparent one
in order to be resourcefully processed [3]. The data miner has
a fundamental role in formulating the problem and preparing
the suitable and relevant data. Additionally, data mining can
turn out non-satisfactory results at first attempts and miners are
to integrate their expertise into the model before starting
another iteration of the process. Moreover, astutely adjusting
some parameters or trying out different algorithms not only
reveals necessary but also requires relevant and consistent
justifications. Particularly, the choice of apposite and pertinent
attributes can grow intractable, intricate and strenuous namely
with large and/or complex classes or datasets. In this sense,
adept miners would simplify this task at an early stage and at
low computational cost by refining and consolidating the raw
data. A typical example would be of David Heckerman [4]
about Hot-Dogs and Barbecue-Sauce false inference.
III. PREDICTING ENGINEERING STUDENTS SUCCESS AND/OR
ENROLLMENT IN MASTERS PROGRAMS
In this paper, we will deal with a particular concern that
considerably affects engineering programs in various types of
Higher Education Institutions. Pondering over the engineering
students primordial performance demonstrated imperative for
an efficient supervision of the attrition rate (on top of probation
and suspension rates) and students chances to further enroll in
Masters studies or to simply achieve (succeed) their
engineering degrees [45] [46] [53] [54]. We will, at a first
stage, aim at discovering the relationship between the most
affecting factors and the aforementioned issues, then, we will
try to construct a predictive model that will endow both
advisors and administrators with a powerful decision-making
tool. The predictive tool or model, as we will call it here, will
help tackle such issues as predicting students GPA, the
attrition rate in the Engineering program, and have an insight of
the enrollment rate in Masters studies. Consequently, it will
decisively help in planning the courseware and designating
needed faculty members, amongst many other pertinent and
resulting matters [16].
Another benefit of this tool is to help advisors and
instructors have an insight about their students, namely the
weak ones. This would help advisors know the capabilities of
their advisees and thus, have a better decision when choosing
their courses during registration. It would help instructors pay
attention to these students during various in-class activities and
team forming. Furthermore, special recommendations could be
prescribed such as doing extra work or having office-hours
visits, etc.
Engineering degrees are mostly offered in two different
curriculum structures. One of them is the 150-credit Bachelor
of Engineering program (BE) and the other one is the BS
(107cr.)/MS (43 cr.) program. In either case, students are to
fulfill strict requirements in order to graduate and hold a degree
in the Engineering profession. Generally, the engineering
program consists of different categories of courses to be
completed by the students to fulfill the graduation
requirements. Engineering students at Notre Dame University-
Louaize (NDU) in Lebanon accounts for approximately 1,200
students (25% of the total number) repartitioned into three
departments and four majors (Electrical, Computer and
Communication, Civil and Environmental, and Mechanical).
Courses are split into four categories: General Education, Core,
Major, and Technical Elective courses. Currently, NDU offers
the Bachelor of Engineering degree but it is also studying the
prospect of launching the BS/MS program. The study
undertaken in this paper applies to both cases since the main
objective is to predict the performance of engineering students
at an early stage of their residency for it affects various factors
related to their academic path such as probation, suspension,
attrition, graduation, and enrollment in further more-advanced
tracks.
The purpose of displaying Fig. 4 below is to first show that
a very strong correlation exists between the performance of a
student in Major courses and his/her cumulative GPA.

Vol. 2, No. 10, 2011
5 | P a g e
Certainly, the latter data is unswervingly related to the
issues stated at the beginning of this paragraph and therefore, it
would be beneficial to use this indicator for it can be obtained,
or as we will see later on, is related to attributes obtained at
early stages of the engineering curricula.
Most of the core courses are usually taken during the first
year. They comprise essentially Math, Physics, and Chemistry
courses. These courses are the prerequisites of almost all
major courses since students are exposed to the fundamental
and basic concepts required to pursue specialized theories on a
later stage. It is straightforward and safe to assume that if a
student has weakly achieved a course, he/she will have fewer
chances to excel or have superior performance in a higher-
level, directly dependent course to which it is a prerequisite.
Indeed, the material covered in the higher-level course
generally relies on the one covered in the prerequisite and is
sometimes a natural/further continuation and advancement in
same or similar concepts and topics. Consequently, core
courses convincingly play a decisive role in the students
performance in major courses.
As a result, the sought predictive tool will be based on the
performance of students in the core-requirement courses, which
can be tracked or depicted early. The study will subtend all
types of students for the purpose of generalization: weak,
average and good performers are included. Table 1 displays
the distribution of students with respect to their overall
performance.
A. Data Preparation
Five hundred Computer and Communication Engineering
students records have been gathered covering a period of
almost seven years. These records consist of comprehensive
transcripts with the students grades in all courses taken
throughout their academic path. These records also include the
number of times students went on probation, those who have
been suspended and the cumulative General Point Average
(GPA) upon graduation.
The records were preprocessed and cleansed in the sense
that records with non-consistent structures were eliminated if
adjustment revealed not possible. In fact, over the period of
seven years, the engineering curricula have slightly changed;
courses have been deleted, modified and/or replaced by new
courses. Additionally, straightforward 4.0 or near 4.0 GPA
have also been discarded in order to avoid misleading the
prediction procedure. This significantly reduced the outlier
presence and noisy data. This part of the data mining process
resulted in 305 clean records with no missing data and seamless
consistency amongst attributes.
Table 1 below displays the distribution of students in
relationship with the cumulative GPA. As mentioned earlier,
various types of learners have been enrolled in this study;
weak, moderate and superior performers have been chosen as
to try enhancing the generalization capacity of the model as
well as it accuracy.
Table 2 shows a snapshot of the students records and
transcripts as obtained from the Registrars Office of NDU.
B. Choosing The Most Pertinent Attributes Using Relieff
Algorithm
The core requirement pool consists of 39 credits comprising
the following courses: CEN 201, ENG 201, ENG 202, MAT
211, MAT 213, MAT 215, MAT 224, MAT 235, MAT 326,
MAT 335, CHM 211, PHS 212, and PHS 213. For more
details regarding a description for each course, please refer to
NDUs online catalog cited in [29]. The core requirement pool
represents a blend of some chemistry, physics, statics, and
mostly math courses, which are mandatory and fundamental.
Furthermore, they introduce the students to the most important
topics and concepts necessary to pursue courses in Electric
Circuits and Electronics, Microprocessor Systems,
Electromagnetism, Signal Processing, Communication,
Programming, Database, Networking, etc.
In order to produce an effective tool, namely with such
database size (not very large), we opted for underlying the most
influential attributes prior to exercising data mining CART.
This helped achieving some sort of pre-pruning before even
training the Decision Tree.
Matlab Relieff algorithm computes the ranks and weights
of attribute (predictors) for an input data matrix and response
vector for classification and regression with K-Nearest
neighbors. When applied to the 305-record data matrix, the
importance of the Math courses outperformed the one of
Physics courses and finally, ENG 201, ENG 202, CHM 211,
and CEN 201 came at the bottom of the list. Particularly, the
following courses were retained for our classification and
regression study based on the outcome of the Relieff algorithm:
MAT 213, MAT 224, MAT 235, MAT 335, PHS 212, and
PHS 213. Furthermore, weights of the latter Math courses
were very close to each other. A similar observation was

Figure 4 - Relationship between Performance in Major courses and the
Cumulative GPA
Table 1 - Students Distribution vs. GPA

Vol. 2, No. 10, 2011
6 | P a g e
depicted for the Physics courses. Consequently, and for other
practical reasons such as keeping a reasonable tree size and
achieving a good balance between the number of attributes and
records, we decided to design a classification tree with two
highly predominant attributes: computed average performance
in Math courses and computed average performance in Physics
courses [43] [51] [52].
It is noteworthy to mention that students grades are ordinal
and they belong to the following set of Letter-Grade/GPA:
A/A
+
(4.0), A
-
(3.7), B
+
(3.3), B (3.0), B
-
(2.7), C
+
(2.3), C
(2.0), C
-
(1.7), D
+
(1.3), D (1.0) and F (0.0).
C. Creating the Model Training, Validating, and Testing
We used Matlab Classification and Regression Trees
(CART) to achieve two objectives. The first one is to elaborate
a regression tree to predict students GPA upon graduation
based on the GPAs they obtained for the abovementioned Math
and Physics courses. The second objective was to create a
binary classification tree that dictates the possibility of a
student to either enroll in an Engineering Masters track or
graduate (succeed) the Bachelor of Engineering (BE) program
since the decision is based on same academic standards and
conditions [17].
For the training stage, 75% of the records were used. The
remaining records were equally distributed between cross-
validation and testing. We also used Matlab Neural Network
Pattern Recognition Tool (NNPRT) to derive the ROC curves
and test the performance of our created model.
D. Neural Networks for Pattern Recognition
Matlab Neural Networks tool for Pattern Recognition is
applied to the 305 students records and Fig. 5 and 6 show
some performance indicators obtained in our case. Seventy
percent of the data have been used for training, 15% for cross-
validation and 15% for testing [19] [41]. In order to apply the
tool to the data, the nominal classes were transformed as
follows:
If the student succeeds the BE or enrolls in Masters, a
1 is assigned to represent the class YES.
If the student did not succeed the BE (failed, dropped
out, or suspended) or did not enroll in Masters (failed
to meet the academic standards), a 0 is assigned to
represent the class NO.
Fig. 5 shows the Error Histogram while deploying the
model; the results are promising and reveal a reliable tool since
the great majority of the error is substantially low.
Fig. 6 shows the ROC curves (Receiver/Relative Operating
Characteristic) obtained. These curves also confirm a highly
performant operating model in the frame of our application.
E. Classification and Regression Trees
Matlab offers an algorithm described by classregtree,
which allows inducing classification and regression trees for
different types of attributes. This algorithm creates a binary
decision tree for predicting the class and/or estimating the
value as a function of the predictors (attributes). Each
branching node within the tree is split based on the values of a
column of the attributes [27]. This algorithm is endowed with
powerful control parameters such a minimum splitting criterion
and a pruning level, which prunes branches giving less
improvement in error cost using a resubstitution method by
turning some branch nodes into leaf nodes and thus, removing
the leaf nodes under the original branch. Furthermore, using
the test function, Matlab allows a 10-fold cross-validation to
compute the cost vector and returns a vector containing the
standard error of each cost value and a scalar value for the best
level of pruning. For a classification tree, the cost of a node is
the sum of the misclassification costs of the observations in that
node whereas for a regression tree, the cost of a node is the
average squared error over the observations in that node. The
classregtree also induces an if-then structure of rules for a
different visualization of the model.
Finally, the eval function of Matlab produces a vector of
predicted response values based on a matrix of new, untrained
data predictors.
Table 2 - Snapshot of the Students' Records

Figure 5 - Error Histogram

Vol. 2, No. 10, 2011
7 | P a g e
Fig. 7 shows how the cumulative GPA is estimated based
on the performance of students in Math and Physics courses
from the core-requirement pool with a best pruning level
reached at 72. It is obvious that the cumulative GPA
significantly worsens when students do not highly perform in
these courses. Nonetheless, it is indubitably not reliable for
accurate prediction because of both the size and type of the
data. On the other hand, it constitutes another indicator that
confirms our postulations and hypotheses.
To use the decision tree effectively and make use of the
results from the Neural Networks Error-Histogram and ROC
curves, a binary classification tree will be deployed and which
will accurately predict the ability of students to either succeed
their bachelor of engineering or enroll in Masters programs.
As shown in Fig. 8, the classification tree designates one class,
YES or NO, to a record based on the attributes. The
former class infers achievement of a BE or enrollment in
Masters, while the latter one signifies that the student would
most probably fail or not be allowed to enroll in Masters, as
previously detailed.
Fig. 8 portrays two different pruning levels: Fig. 8-b
exhibits the best-pruning level while Fig. 8-a exhibits a pruning
level that is less by one than the best level. The purpose is to
have a deeper and wider bifurcation when a student presents
close attributes with respect to some deciding node values and
has to be evaluated by the model.
Figure 8.a - Classification Tree (Best Level minus one)
The induced if-then structure as given by Matlab is shown
below. This structure is related to the decision tree obtained for
the best pruning level.

After examining the classification tree, we can summarize
the following results:
If a students GPA in Math courses is above 2.7375,
then he/she is most likely to succeed and/or enroll in
Masters.
If a students GPA in Math courses is above 2.0125,
then the condition for a positive outcome is to have a
GPA on the Physics courses of at least 2.75.
Otherwise, the chances for students who have a GPA
in Math courses less than 2.0125 are scarce to enroll
in Masters or accomplish a Bachelor of Engineering.
IV. RESULTS ANALYSIS, CONCLUSION
In this paper, we carried out a study to find a reasonably
accurate and reliable predictive tool that enables academicians
(instructors and advisors) and administrators to decide about
the enrollment of engineering students in Masters studies or to
succeed a Bachelor-of-Engineering program.
The study has been conducted in different stages and on
different levels. The strong correlation between students
performance in major courses, which are usually taken during
the last three years of a Bachelor-of-Engineering program or
during the Masters curriculum, and their cumulative GPA was
Decision tree for classification
1 if MAT_GPA<2.7375 then node 2 else node 3
2 if MAT_GPA<2.0125 then node 4 else node 5
3 class = YES
4 class = NO
5 if PHS_GPA<2.75 then node 6 else node 7
6 class = NO
7 class = YES

Figure 7 - Regression Tree for Estimation of the Cumulative GPA

Figure 6 - ROC Curves

Prediction of Masters Enrollment/BE-Success
(One Pruning Level below Best Level)
Vol. 2, No. 10, 2011
8 | P a g e
demonstrated. The objective of this part of the study was to
enable linking core-requirement courses with the GPA and
thus, allow predicting students overall performance at an early
stage of their studies.
At first, Matlab Neural Networks Pattern Recognition tool
was applied in order to examine and decide of the accuracy of
the predictive tool, which was to be eventually developed. The
Error Histogram as well as the ROC curves demonstrated high
level of satisfaction and reliability, namely given the size of the
data.
Preprocessing the data was of high importance because it
helped achieve a low level of outliers and present the data in a
more efficient way to classification and regression trees.
Particularly, the number of attributes was relatively high when
compared to the number of records and thus, preliminary
studies have been conducted and which efficiently, and without
loss of information, reduced the size of attributes to two highly
decisive ones. Matlab Relieff function was used to reveal the
most influential attributes to be taken into account.
At the last stage, Matlab classregtree tool was used in two
steps. The first one was to create a regression tree that
estimates the cumulative GPA based on the attributes. This
gave us another confirmation of the postulation we started from
and then, at a later step, a binary classification tree was
achieved after cross-validation and appropriate pruning. This
decision tree created an if-then rule structure that enables the
engineering staff to ponder over students chances of
succeeding their engineering studies.
The results revealed promising especially when discussed
with Math and Physics courses instructors and engineering
advisors who all agreed on the discovered/learnt liaison and
patterns. It confirmed that students, who are weak performers
particularly in this pool of courses, exhibit difficulties in most
of the cases in comprehending advanced engineering concepts
and in achieving high performance in major courses.
Furthermore, the study gave specific numbers and thresholds in
courses taken at early stages that would alarm academicians
about the situation of the concerned students.
Finally, this study will indubitably help in predicting the
number of students who will reach the end of the engineering
program and thus, constitutes a performant tool for decision-
making and forecasting enrollment and courseware planning as
well as pondering over attrition in engineering studies. It
would also endow advisors and courses instructors an
anticipated estimation of their advisees and students
capabilities during registration and in-class activities and a
reliable platform for special recommendations.
REFERENCES
[1] R.S.J.d. Baker, Data Mining for Education, In International
Encyclopedia of Education, vol. 7, B. McGaw, P. Peterson, E. Baker
(Eds.), 3e, Oxford, UK: Elsevier, 2010, pp. 112-118.
[2] http://www.educationaldatamining.org/
[3] J. Teresko, Information Rich, Knowledge Poor?, Data Warehouses
Transform Information into Competitive Intelligence, February 3, 1999.
(http://www.industryweek.com/articles/information_rich_knowledge_po
or_245.aspx)
[4] C. Silverstein, S. Brin, R. Motwani, and J. Ullman, Scalable
Techniques for Mining Causal Structures, Data Mining and Knowledge
Discovery, volume 4, numbers 2-3, pp. 163-192, 2000, DOI:
10.1023/A:1009891813863
[5] P. Domingos, Toward knowledge-rich data mining, In Proceedings of
Data Min. Knowl. Discov., 2007, volume 15, issue 1, pp. 21-28, DOI:
10.1007/s10618-007-0069-7
[6] http://www.kdnuggets.com/data_mining_course/x1-intro-to-data-
mining-notes.html (retrieved in September 2011)
[7] http://www.focus.com/fyi/10-largest-databases-in-the-world/
[8] http://beyondrelational.com/justlearned/posts/290/top-10-largest-
databases-in-the-world.aspx
[9] M. Berry and G. Linoff, Data Mining Technique: For Marketing, Sales,
and Customer Support, New York: Wiley Computer Publishing, 1997
[10] U. M. Fayyad, G. Piaetsky-Shapiro, and P. Smyth, From Data Mining
to Knowledge Discovery: An Overview, In Advances In Knowledge
Discovery and Data Mining, AAAI/MIT press, Cambridge mass, 1996.
[11] I. H. Witten, E. Frank, and M. A. Hall, Data Mining: Practical Machine
Learning Tools and Techniques, Third Edition, Morgan Kaufmann,
2011.
[12] Jiawei Han, Micheline Kamber, and Jian Pei, Data Mining: Concepts
and Techniques, Third Edition, Morgan Kaufmann, 2011.
[13] Krzysztof J. Cios, Witold Pedrycz, and Roman W. Swiniarski, Data
Mining: A Knowledge Discovery Approach, Springer 2007.
[14] D. J. Hand, Heikki Mannila, and Padhraic Smyth, Principles of Data
Mining, Massachusetts Institute of Technology, 2001.
[15] N. Delavari, S. Phon-Amnuaisuk, and M. R. Beikzadeh, Data Mining
Application in Higher Learning Institutions, Informatics in Education,
vo. 7, no. 1, pp. 31-54, 2008.
[16] S. Sembiring, M. Zarlis, D. Hartama, S. Ramliana, and E. Wani,
Prediction of Student Academic Performance by an Application of Data
Mining Techniques, International Conference on Management and
Artificial Intelligence, IACSIT Press, IPEDR vol. 6, pp. 110-114, 2011.
[17] S. A. Kumar and M. N. Vijayalakshmi, Efficiency of Decision Trees in
Predicting Student's Academic Performance, First International
Conference on Computer Science, Engineering and Applications, CS
and IT 02, pp. 335-343, 2011.
[18] C. Ho Yu, S. DiGangi, A. Jannasch-Pennell and C. Kaprolet, A Data
Mining Approach for Identifying Predictors of Student Retention from
Sophomore to Junior Year, Journal of Data Science, vol. 8, pp. 307-
325, 2010. (neural networks, cross validation)
[19] Fausett, Laurene (1994), Fundamentals of Neural Networks:
Architectures, Algorithms and Applications, Prentice-Hall, New Jersey,
USA.

Figure 8.b - Classification Tree (Best Pruning Level)
Vol. 2, No. 10, 2011
9 | P a g e
[20] L. Getoor, Linking Mining: A New Data Mining Challenge, ACM
SIGKDD Explorations, vol. 5, no. 1, pp. 84-89, 2003. (large data)
[21] D. Denning, An Intrusion Detection Model, Proceedings of the
Seventh IEEE Symposium on Security and Privacy, pp. 119-131, May
1986.
[22] M. R Smith and T. Martinez, Improving Classification Accuracy by
Identifying and Removing Instances that Should Be Misclassified,
Proceedings of International Joint Conference on Neural Networks,
pp. 2690-2697, 2011.
[23] R. Agrawal, T. Imielinski, and A. Swami, Mining Association Rules
Between Sets of Items in Large Databases, Proceedings of the ACM
SIGMOD International Conference on Management of Data, pp. 207-
216, May 1993.
[24] G. Piatetsky-Shapiro, Discovery, analysis, and presentation of strong
rules, in G. Piatetsky-Shapiro and W. J. Frawley, eds, Knowledge
Discovery in Databases, AAAI/MIT Press, Cambridge, MA, 1991.
[25] R. Nock and F. Nielsen, On Weighting Clustering, IEEE Trans. on
Pattern Analysis and Machine Intelligence, vol. 28, no. 8, pp. 1-13,
August 2006.
[26] J.H.Ward Jr., Hierarchical Grouping to Optimize an Objective
Function, Journal of the American Statistical Association, vol. 48, pp.
236-244, 1963.
[27] J. R. Quinlan, Induction of Decision Trees, Machine Learning, vol. 1,
Kluwer Academic Publishers, Boston, pp. 81-106, 1986.
[28] http://www.educationaldatamining.org/JEDM/
[29] http://www.ndu.edu.lb/administration/registrar/catalogs.htm
[30] R. Agrawal and R. Srikant, Fast Algorithms for Mining Association
Rules in Large Databases, in VLDB94 Proceedings of the 20
th

International Conference on Very Large Data Bases, Morgan Kaufmann
Publishers Inc. San Francisco, CA, USA, 1994. (a priori)
[31] P. Fournier-Viger, U. Faghihi, R. Nkambou, and E. M. Nguifo,
CMRULES: An Efficient Algorithm for Mining Sequential Rules
Common to Several Sequences, in Proceedings of the Twenty-Third
International Florida Artificial Intelligence Research Society Conference
(FLAIRS 2010), AAAI Publications, pp. 410-415, 2010. (association
task)
[32] A. Nandeshwar, T. Menzies, and A. Nelson, Learning Patterns of
University Student Retention, Expert Systems with Applications, vol.
38, issue 12, pp. 14984-14996, Nov.-Dec. 2011. (attrition)
[33] R. Nock and F. Nielsen, On Weighting Clustering, IEEE Transactions
on Pattern Analysis and Machine Intelligence, vol. 28, no. 8, pp. 1-13,
August 2006.
[34] M. A. Maalouf, Machine Learning and Data Mining for Computer
Security: Methods and Applications, Springer-Verlag, Limited 2006.
(outlier)
[35] J. Dean and S. Ghemawat, MapReduce: simplified data processing on
large clusters, in OSDI'04 Proceedings of the 6th conference on
Symposium on Opearting Systems Design & Implementation, USENIX
Association Berkeley, CA, USA, vol. 6, 2004. (large data)
[36] J. Luan, An Exploratory Approach to Data Mining in Higher
Education: A Primer and a Case Study, Paper presented at the AIR
Forum, Seattle, Wash., 2000. (Higher Education)
[37] R.S. Baker, A. T. Corbett, K. R. Koedinger, and A. Z. Wagner, Off-
Task Behavior in the Cognitive Tutor Classroom: When Students Game
The System, in Proceedings of ACM CHI: Computer-Human
Interaction, pp. 383-390, 2004. (higher education)
[38] R.S.J.d. Baker, S. K. D'Mello, M. M. T. Rodrigo, and A. C. Graesser,
Better to Be Frustrated than Bored: The Incidence, Persistence, and
Impact of Learners' Cognitive-Affective States during Interactions with
Three Different Computer-Based Learning Environments, International
Journal of Human-Computer Studies, vol. 68, no. 4, pp. 223-241, 2010.
(higher education)
[39] R.S.J.d. Baker, and K. Yacef, The State of Educational Data Mining in
2009: A Review and Future Visions, Journal of Educational Data
Mining, vol. 1, issue 1, pp. 3-17, 2009.
[40] F. Thabtah, Pruning Techniques in Associative Classification: Survey
and Comparison, Journal of Digital Information Management, vol. 4,
no. 3, pp. 197-202, September 2006. (pruning)
[41] Sarle, Warren S. (1994), Neural Networks and Statistical Models,
Proceedings of the Nineteenth Annual SAS Users Group International
Conference, April, pp 1-13.
[42] J. Quinlan, Simplifying Decision Trees, International Journal of Man-
Machine Studies, vol. 27, no. 3, pp. 221-248, 1987. (Decision Trees)
[43] L.-X. Zhang, J.-X. Wang, Y.-N. Zhao, and Z.-H. Yang, A Novel
Hybrid Feature Selection Algorithm: Using Relieff Estimation for GA-
Wrapper Search, in Proceedings of the Second International Conference
on Machine Learning and Cybernetics, pp. 380-384, 2-5 November
2003.
[44] P. Shekhawat and S. Dhande, A Classification Technique using
Associative Classification, International Journal of Computer
Applications, vol. 20, no.5, pp. 20-28, April 2011. (AC)
[45] G. P. Adanez and A. D. Velasco, Predicting Academic Success of
Engineering Students in Technical Drawing from Visualization Test
Scores, Journal for Geometry and Graphics, vol. 6, no. 1, pp. 99-109,
2002. (engineering)
[46] V.O. Oladokun, A.T. Adebanjo, and O.E. Charles-Owaba, Predicting
Students Academic Performance using Artificial Neural Network: A
Case Study of an Engineering Course, The Pacific Journal of Science
and Technology, vol. 9, no. 1, pp. 72-79, 2008. (engineering)
[47] F. Thabtah, P. Cowling, and Y. Peng, Multiple Label Classification
Rules Approach, Journal of Knowledge and Information System, vol. 9,
pp. 109-129, Springer-Verlag, 2006. (large data)
[48] T.-M. Huang, V. Kecman, and I. Kopriva, Kernel Based Algorithms for
Mining Huge Data Sets: Supervised, Semi-supervised, and Unsupervised
Learning, Springer-Verlag Berlin Heidelberg, 2006. (large data +
supervised)
[49] J. Meenakumari and R. Krishnaveni, Transforming Higher educational
institution administration through ICT, (IJACSA) International Journal
of Advanced Computer Science and Applications, vol. 2, no. 8, pp. 51-
54, 2011. (higher education as a business)
[50] V. Kumar and A. Chadha, An Empirical Study of the Applications of
Data Mining Techniques in Higher Education, (IJACSA) International
Journal of Advanced Computer Science and Applications, vol. 2, no. 3,
pp. 80-84, March 2011.
[51] Y. Wang and F. Makedon, Application of Relief-F Feature Filtering
Algorithm to Selecting Informative Genes for Cancer Classification
Using Microarray Data, in Proceedings of the 2004 IEEE
Computational Systems Bioinformatics Conference, IEEE Computer
Society Washington, DC, USA, 2004
http://dx.doi.org/10.1109/CSB.2004.35 (relieff)
[52] I. Kononenko, E. Simec, and M. R. Sikonja, Overcoming the Myopia of
Inductive Learning Algorithms with Relieff, Journal of Applied
Intelligence, vol. 7, no. 1, pp. 39-55, 1997. (relieff)
[53] A. H. Basha, A. Govardhan, S. V. Raju, and N. Sultana, A Comparative
Analysis of Prediction Techniques for Predicting Graduate Rate of
University, European Journal of Scientific Research, vo. 46, no. 2, pp.
186-193, 2010.
[54] V. Ramesh, P. Parkavi, and P. Yasodha, Performance Analysis of Data
Mining Techniques for Placement Chance Prediction, International
Journal of Scientific and Engineering Research, vol. 2, issue 8, pp. 1-6,
August 2011.
AUTHORS PROFILE
Dr. Chady El Moucary graduated from the Lebanese University
Faculty of Engineering with a diploma in Electrical and Electronics
Engineering. He pursued his postgraduate studies in Paris (France) with a
scholarship from the French National Center for Scientific Research (CNRS)
and received his PhD degree in Electrical Engineering from the University of
Paris XI and the SUPELEC in 2000. Currently, Dr. El Moucary is a full-time
Assistant Professor and Researcher at Notre Dame University -Louaize (NDU,
Lebanon) and the Coordinator of the Faculty of Engineering in NDUs North
Lebanon Campus. His research interests and publications are in Electric
Machine Control, Digital Watermarking, and Data Mining. He is also a
member of the Scientific Committee of the Order of Engineers in Tripoli
(Lebanon) and a reviewer for diverse reputable worldwide conferences as well
as an active member in many other educational/academic committees.
Vol. 2, No. 10, 2011

10 | P a g e
Model of Temperature Dependence Shape of
Ytterbium -doped Fiber Amplifier Operating at 915
nm Pumping Configuration

Abdel Hakeim M. Husein
a*
,
a*
Department of Physics, Al-Aqsa University, P.O. Box
4051, Gaza, Gaza Strip, Palestine
Fady I. EL-Nahal
b

b
Department of Elec. Eng., Islamic University of Gaza,
Gaza, Gaza Strip, Palestine

Abstract We numerically analyze the temperature dependence
of an ytterbium-doped fiber amplifier (YDFA) operating at 915
nm, investigating its gain and Noise Figure properties variation
with temperature. The temperature-dependent gain and noise
figure variation with YDFA length are numerically obtained for
the temperature range of +20
0
C to +70
0
C. The results show that
good intrinsic output stability against temperature change can be
achieved in ytterbium doped fiber amplifiers even when
operating at high gain regime with small signal input. This result
demonstrates the great potential for stable high power laser
communication systems based on ytterbium system.

Keywords- YDFA; Gain; Noise Figure.
I. INTRODUCTION
Fiber lasers and amplifiers have attracted great interest
recently, because they offer the advantages of compact size,
high gain, guided mode propagation, better stability and their
outstanding thermo-optical properties [1-4]. Ytterbium (Yb
3+
) -
doped fiber Amplifier (YDFA) has a great potential because it
does not have some of the drawbacks associated with erbium-
doped amplifier: excited state absorption phenomenon that can
reduce the pump efficiency and concentration quenching by
interionic energy transfer do not occur, and high doping levels
are possible. Thus, it offers high output power (or gain) with a
smaller fiber length. YDFAs have a simple energy level
structure and provide amplification over a broad wavelength
range from 975 to 1200 nm. Moreover, YDFAs can offer high
output power and excellent power conversion efficiency [1,5-
12].
YDFAs have great potential in many applications,
including power amplification, sensing applications, free-space
laser communications, and chirped-pulse amplification of ultra-
short pulses [1,12-14].
This paper explores the effects of temperature on amplifier
performance. For an amplifier, the temperature affects relative
extraction by the signal and ASE. Thermal management is a
critical issue and cannot be ignored. Some papers about
temperature effect of Er
3+
-doped fiber laser and amplifier were
reported [15-17]. Temperature can affect the absorption and
emission cross sections [2]. This paper demonstrates output
characteristics of Yb
3+
-doped fiber laser at different
temperatures degrees. So from this research the temperature
can affect the gain and noise figure (NF) at different length.
Established methods of modeling erbium amplifiers can be
used to model ytterbium system [11,18-19]. However,
modeling temperature sensitivity is completely different. The
small energy gaps in the relevant stark levels in erbium
systems, makes the determination of distinct sub-transition
characteristics quite difficult [17], while this is not true in
ytterbium system. Accurate characterization of Yb
3+
absorption
and emission cross sections is crucial [20,21].
II. THEORETICAL MODEL
We used standard rate equations for two-level systems to
describe the gain and propagation characteristics of the Yb-
doped ber amplier operating at 975 nm because the ASE
power is negligible for a high power amplier with sufcient
input signal (about 1 mW). After the overlap factors are
introduced and the ber loss ignored, the simplied two-level
rate equations and propagation equations are given as follows
[12]:
2 2
1 2 1 2
( , )
[ ] [ ]
p p
s s
sa se pa pe
s p
P
dN z t P N
N N N N
dt Ah Ah
o o o o
u u t
I
I
= + (1)
1 2 0
( , ) ( , ) N z t N z t N = (2)
2 1
( )
[ ( , ) ( , ) ]
s
s s se sa
dP z
P N z t N z t
dz
o o = I (3)
2 1
( )
[ ( , ) ( , ) ]
p
p p pe pa
dP z
P N z t N z t
dz
o o = I (4)
Here, N
0
is the Yb-dopant concentration, N
1
and N
2
are the
ground and upper-level populations.
s
I and
p
I are the
oveperlapping factor between the pump (signal) and the ber-
doped area. ( , ), ( , )
s p
P z t P z t are the signal and pump power
respectively.
sa se
and o o are the signal absorption and
emission cross sections.
pa pe
and o o are the signal
absorption and emission cross sections. and
s p
u u are the
Vol. 2, No. 10, 2011

11 | P a g e
frequencies of signal and pump light, respectively. A is the
doped area of the ber. t is the upper state lifetime.
Under the condition of the steady state regime, where all of
the level population are time invariant i.e.,
( , )
0 ( 1, 2),
i
dN z t
i
dt
= =
2
1 2 1 2
[ ] [ ]
p p
s s
sa se pa pe
s p
P
N P
N N N N
Ah Ah
o o o o
t u u
I
I
= + (5)

1 0 2
( , ) ( , ) ( , ) N z t N z t N z t = (6)
A. Spectroscopy of ytterbium in silica
Ytterbium in silica is a simple, two level system having
four Stark levels in the lower manifold F
7/2
and three Stark
levels in upper manifold F
5/2
. An energy level diagram specific
to Nufern 5/125 fiber is shown in Figure 1 [22]. Because the
splitting of the levels depends on the glass composition,
concentration of dopants and co-dopants, and the degree of
structure disorder of the glass network, the energy level
diagram for Yb in silica may vary with each individual fiber.
The absorption and emission cross-sections for Yb in silica are
related to the temperature and the energy of the levels by the
following relationships:

Figure 1: Energy level diagram of Yb in silica with 976 nm, 1040 nm, and 1064
nm transitions labeled [22].
( , ) ( )
x
x
E
g d KT
sa
sa xy E
d
x a y e
KT
x a
e
T
e
o u o u
= =
=
=
(7)
( , ) ( )
x
x
E
g d KT
se
se xy E g
x e y a
KT
x e
e
T
e
o u o u
= =
=
=
(8)
where E
x
are the energies of the level difference between
level x and a, when xe(a,d) where a is the ground state in
lower manifold and the energies of the level difference
between levels x and e, in the ground state in upper manifold
when xe(e,g), T is the temperature, K=
23
1.3806 10 / J k
is
Boltzmann's constant, and ( )
sa
xy
o u and ( )
se
xy
o u are the
absorption and emission cross-sections of the sub-transitions
[17]. To be able to calculated
se
o , it needed to use McCumber
theory [23] in the Yb system, this expressed as
( )/
( , )
h KT
se sa
T e
c u
o u o

= (9)
where
/ / /
/ /
xy / /
1
, E
1
ba ca da
ge fe
E KT E KT E KT
KT Eea KT
x y E KT E KT
e e e
e e E E
e e
c

+ + +
= =
+ +
(10)
And h=6.626 x 10
-34
J.s. is Planck constant and c is an
active energy and given by
1
2
exp( / )
N
KT
N
c = (11)
Where N
1
and N
2
are Yb doping concentration at the upper
and lower energy levels, respectively. Using the absorption and
emission cross section from Eqs.(7-11), and the propagation
equations (3) and (4) can even be solved analytically in this
case [23].
The small signal gain G for active length of the fiber L can
be given by
| |
2 1
( ) exp ( ( , ) ( , ) )
s se sa
G N z t N z t L o o = I (12)
Figure 2 and 3 show the variation of Yb absorption and
emission cross sections at various temperatures respectively
[21]. It is clear from figure 3 that the signal absorption cross
section declines with increasing wavelength and it is almost 0
at 1064 nm. The term
1
( , )
sa
N z t o can be ignored as it is a
decreasing function of signal wavelength, then
| |
10 2
( ) 10log exp
s se
G dB N L o = I (13)
The amplified spontaneous emission (ASE) noise spectrum
uses the noise figure (NF) given or as input parameter. In the
practical case the ASE is presented at input of the doped fiber,
therefore the amplified input ASE
i
ASE
P spectral density can be
added to the amplified output ASE spectral density
o
ASE
P
, so

o i
ASE amp ASE
P P P G = + (14)
Where G is the gain and
amp
P is the spectral density of ASE
generated by the doped fiber. For input ASE gives the signal
spontaneous beat noise (1/G) limited noise figure as a function
of the signal gain and input and output ASE spectral densities
therefore the NF can be expressed as:
( ) ( ) 1
o i
ASE s ASE s
s s
P P
NF
G Gh h

u u
= + (15)
Where 1/G is the beat noise, ( )
o
ASE s
P is the output ASE
spectral density (Watt/Hertz) at signal wavelength, ( )
i
ASE s
P is
the input ASE spectral density at signal wavelength, and
s
u is
the frequency of the signal wavelength. For lower gains, a
much simpler model which ignores the effect of ASE on the
level populations can be used, so the ( ) 0
i
ASE s
P = .
For each signal wavelength the noise figure can be
calculated in decibel (dB) and is given by:
Vol. 2, No. 10, 2011

12 | P a g e
10
( ) 1
( ) 10 log
o
ASE s
s
P
NF dB
G Gh
u
| |
| = +
|
\ .
(16)

Figure 2: Yb absorption cross sections at various temperatures. [18]

Figure 3: Yb emission cross sections at various temperatures [21].
III. RESULTS AND DISSCUSSIONS
For numerical calculation, the fiber parameters for YDFA
amplifier are shown in Table 1. We studied the variation of
gain and NF with the length of the amplifier over the
temperature range from 20
o
C to 70
o
C at signal wavelength of
1064 nm. The results are shown in Figures 4 and 5,
respectively. It is clear from the results that the signal gain
raises with increases in the length, at the same time the gain
declines when the temperature increases. However the NF
increases when the temperature rises. Furthermore the NF
increases with increasing the length. It is clear from the results
that the variation of the NF with temperature is minimal for
lengths over 5 m.
IV. CONCLUSION
A YDFA model has been introduced including the
temperature effects for gain and noise figure of a length of the
YDFA amplifier. The temperature dependence of the gain and
noise figure on various temperatures was taken into
consideration which shows that the performance of YDFA
exhibit excellent low temperature sensitivity. The analytical
solution of the propagation equations has also been derived for
the temperature range from 20
0
C, to +70
0
C for finding the
gain. These results demonstrate that highly stable ytterbium
doped fiber amplifiers can potentially be achieved.
Table 1: The ber parameters and symbols used in the
numerical calculations [18]
Value Definitions Symbol
27 2
2.3 10 m

Signal absorption
cross sections
sa
o
27 2
1.09 10 m

Signal emission
cross sections
se
o
14
3.279 10 Hz

Frequency of input
signal
s
u
14
2.819 10 Hz

Frequency of pump
signal
p
u
1064nm Signal Wavelength
s

915nm Pump wavelength
p

0.84 ms The upper state
lifetime
t
25 3
3.35 10 m

Yb-dopant
concentration
0
N
11 2
7.8 10 m

Doped area of the
fiber
A
0.6 Oveperlapping
factor between the signal
and the ber-doped area
s
I
0.01 Oveperlapping
factor between the pump
and the ber-doped area
p
I
10 to 10 W W

Input Pump power
of 20
0
C and 70
0
C
p
P
0 Amplified input
power of ASE
( 0)
i
ASE
P L =
0 to7m Length of the fiber L

Vol. 2, No. 10, 2011

13 | P a g e

Figure 4: The change of the signal gain with temperature and length.

Figure 5: The change of the Noise Figure with temperature and length.
REFERENCES
[1] Rudiger Paschotta, Johan Nilsson, Anne C. Tropper, and David C.
Hanna, Ytterbium-Doped Fiber Amplifiers, IEEE JOURNAL OF
QUANTUM ELECTRONICS, vol. 33, no. 7, JULY 1997.
[2] J. Chen, Z. Sui, F. Chen and J. Wang, Output characteristics of Yb
3+
-
doped fiber laser at different temperatures, vol. 4, no. 3, Chin. Opt. Lett.
(2006).
[3] D. Xue, Q. Lou, J. Zhou, L. Kony, J. Li, Chin. Opt. Lett. 3,345 (2005).
[4] M.J.F. Digonnet, Rare-Earth-Doped Fiber Lasers and Amplifiers model
for rare-earth-doped ber ampliers and lasers, CRC Press; 2nd edition
(2001).
[5] H. M. Pask, R.J. Carman, D. C. Hanna, A. C. Tropper, C. J.
Mackechinc, P. R. Barber, and J. M. Dawes, IEEE Sel. Top. Quantum
Electron. 1, 2 (1995).
[6] D. C. Hanna, R. M. Percival, I. R. Perry, R. G. Smart, P. J. Suni, J. E.
Townsend, and A. C. Tropper, Continuous-wave oscillation of a
monomode ytterbium-doped fiber laser, Electron. Lett., vol. 24,
pp.11111113, (1988).
[7] D. C. Hanna, R. M. Percival, I. R. Perry, R. G. Smart, P. J. Suni, and A.
C. Tropper, An ytterbium-doped monomode fiber laser: broadband
tunable operation from 1.010 _m to 1.162 _m and three-level operation
at 974 nm, J. Mod. Opt., vol. 37, pp. 517525, (1990).
[8] J. C. Mackechnie, W. L. Barnes, D. C. Hanna, and J. E. Townsend,
High power ytterbium (Yb3+)-doped fiber laser operating in the 1.12
_m region, Electron. Lett., vol. 29, pp. 5253, (1993).
[9] J. Y. Allain, M. Monerie, H. Poignant, and T. Georges, High-efficiency
ytterbium-doped fluoride fiber laser, J. Non-Crystalline Solids, vol.
161, pp. 270273, (1993).
[10] S. Magne, M. Druetta, J. P. Goure, J. C. Thevenin, P. Ferdinand, and G.
Monnom, An ytterbium-doped monomode fiber laser: amplified
spontaneous emission, modeling of the gain and tunability in an external
cavity, J. Lumin., vol. 60, pp. 647650, (1994).
[11] H. M. Pask, R. J. Carman, D. C. Hanna, A. C. Tropper, C. J.
Mackechnie, P. R. arber, and J. M. Dawes, Ytterbium-doped silica
fiber lasers: versatile sources for the 11.2 _m region, IEEE J. Select.
Topics Quantum Electron., vol. 1, pp. 213, (1995).
[12] Liu Yan_, Wang Chunyu, Lu Yutian, A four-passed ytterbium-doped
fiber amplifier, Optics & Laser Technology, pp. 1111-1114 39 (2007).
[13] R. Paschotta, D. C. Hanna, P. DeNatale, G. Modugno, M. Inguscio, and
P. Laporta, Power amplifier for 1083 nm using ytterbium doped fiber,
Opt. Commun., vol. 136, pp. 243246, (1997).
[14] V. Cautaerts, D. J. Richardson, R. Paschotta, and D. C. Hanna,
Stretched pulse Yb3+: silica fiber laser, Opt. Lett., vol. 22, no. 5, pp.
316-318 (1997).
[15] N. Kagi, A. Oyobe, and K. Nakamura, J. Lightwave Technol. 9, 261
(1991).
[16] H. Tobbon, Electron. Lett. 29, 667 (1993).
[17] M. Bolshtyansky, P. Wysocki and N. Conti, Model of temperature
dependence for gain shape of erbium-doped fiber amplifier, Journal of
Lightwave Technology, vol.18, 1533-1540 (2000).
[18] C.R. Giles, E. Desurvire, Modeling erbium-doped fiber amplifiers, J.
Lightwave Technol., vol. 19, no. 7, pp. 271 283, (1991).
[19] Yu. A. Varaksa, G. V. Sinitsyn, and M. A. Khodasevich, Modeling the
gain and amplified spontaneous emission spectra of erbium-doped
fiber amplifiers, J. Appl. Spectrosc., vol. 73, no. 2, pp. 309312,
(2006).
[20] Xiang Peng, Joseph McLaughlin and Liang Dong, "Temperature
Dependence of Ytterbium Doped Fiber Amplifiers", Conference Paper
Optical Amplifiers and Their Applications (OAA) Budapest, Hungary,
August 7, (2005).
[21] Xiang Peng and Liang Dong, "Temperature dependence of ytterbium-
doped fiber amplifiers", J.Opt.Soc. Am. B, Vol. 25, Issue 1, pp. 126-130
(2008).
[22] Leanne J. Henry, Thomas M. Shay, Dane W. Hult and Ken B. Rowland
Jr "Thermal effects in narrow linewidth single and two tone fiber lasers",
vol. 19, no. 7 Optics express 6165 (2011)
[23] C. Barnard, P. Myslinski, J. Chrostowski, and M. Kavehrad, Analytical
model for rare-earth-doped ber ampliers and lasers, IEEE J.
Quantum Electron., vol. 30, pp. 18171830, (1994).
Vol. 2, No. 10, 2011
14 | P a g e
Healthcare Providers Perceptions towards Health
Information Applications at King Abdul-Aziz
Medical City, Saudi Arabia

Abeer Al-Harbi
Health & Hospital Administration department, College of Business Administration
King Saud University (KSU)
Riyadh, Saudi Arabia
Abstract-
The purpose of this study was to assess the perceptions of
healthcare providers towards health information technology
applications in King Abdul-Aziz Medical City in terms of
benefits, barriers, and motivations.
The study population consists of all healthcare providers working
at KAMC. A sample size of 623 was drawn from a population of
7493 healthcare providers using convenience random sampling
method. 377 were returned, giving a response rate of 60.5%.
A self-administered questionnaire was developed based on
extended literature review and comprised 25 statements on a five-
point Likert-scale.
Results indicate that the majority of healthcare providers use
KAMC health information applications. The majority of
healthcare providers perceived that the applications are valuable
and beneficial. However, healthcare providers were split over the
barriers to HIT use in KAMC. As for drivers, healthcare
providers generally would be motivated to use the IT applications
by provision of new applications and training, contribution in
change hospital's work procedures, and provision of technical
support. Also, there were many barriers identified by healthcare
providers. These were insufficient number of computers,
frequent system down, and the use of computerized systems is
time consuming. Finally, there were significant differences in the
perceptions with respect to gender, occupation, and training.

Keywords- Healthcare providers; Health Information Technology;
Computerized Patient Record; King Abdul-Aziz Medical City.
I. INTRODUCTION
a. Health Information Technology (HIT):
Healthcare information technology (HIT) has become a key
preoccupation of healthcare systems worldwide [1, 2]. A
review of the literature reveals that there is significant
consensus that the implementation of electronic health records
(EHRs) and HIT systems is considered among the highest
priorities of modern healthcare systems [3].
Clinical practices rely heavily on the collection and analysis
of medical data for decision-making abilities when caring for
patients [4]. Thus, health information systems are capable of
having a significant, positive impact on patient care within
healthcare settings [5].
Health information technology is in general increasingly
viewed as the most promising tool for improving the overall
quality, safety and efficiency of the health delivery system [6,7,
8]. The institute of Medicine (IOM) identified information
technology as one the critical forces that could significantly
improve healthcare quality and safety [9].
One of the most challenging areas of health information
technology is integrating it into the workflow of the healthcare
providers [10]. Despite the increasing availability of health
information technology applications, anecdotal evidence
suggests that its use has not been well accepted by healthcare
providers [11,12]. Acceptance of information technologies has
occupied a central role in information technology research.
There have been many studies investigating IT acceptance in
different settings at both individual and organizational levels of
analysis and different theoretical models have been used
[13,14]. The literature provides evidence of failed clinical
system implementations, due to lack of adoption by users [15].
However, with few significant exceptions, information system
research is scarce regarding information technology acceptance
in a healthcare environment [16,17,18,19].
In Saudi Arabia, the government strives to improve quality
and safety of healthcare services through adoption health
information technology [20]. However, most Saudi health
organizations have no electronic health records (HER) systems
implemented in their facilities, and they are totally dependent
either on manual paper work or on very basic software tools to
do their day to day tasks such as patient admissions [21].
KAMC is one of the few hospitals that have a basic EHR
system which was later replaced by a computerized Patient
Record (CPR) system. CPR is a single integrated system with a
comprehensive suite of modules that provides depth and
breadth of patient-care support and workflow management.
CPR system streamlines administrative functions and
eliminates paperwork to get caregivers back in the business of
quality patient care. CPR system provides for an array of
technological imperatives, including Computerized Physician
Order Entry (CPOE), Clinical Decision Support (CDS),
automated nursing documentation, integrated pharmacy and
automated medication administration. A CPOE system, for
example, makes prescription orders legible, identifies the
correct medication and dose as well as signals alerts for
Vol. 2, No. 10, 2011
15 | P a g e
potential medication interactions or allergic reactions [20].
According to Dr. David Brailer, cited in Harrison and Daly
[22], CPOE reduces medication errors by 20 percent .
Despite the importance of HIT in improving healthcare
efficiency, there were few studies carried on use, barriers and
drivers to HIT in Saudi health organizations. Therefore, there is
a need for investigating the perceptions of healthcare providers
towards the health information technology applications. This
research is an attempt to understand the perceptions of
benefits, barriers, and motivation toward the use of health
information applications. In addition, the research will
investigate the effect of demographic and organizational
variables on the perceptions of the healthcare providers towards
the health information technology applications.
b. King Abdul-Aziz Medical City (KAMC):
King Abdul-Aziz Medical City commenced its operations
in 1983 in Riyadh under National Guard Health Affairs
(NGHA). NGHA has passed the requirements for accreditation
under the (JCI) Joint Commission International standards with
excellent performance in December 2009. The total bed
capacity of the hospital is 847 beds. The average length of stay
is 4.6 days, and the average number of outpatient visits per day
is 3,145 patients. Total number of physicians is 1564, total
number of nurses is 3921, and the total number of
clinical/paramedical staff is 2008.
II. METHODS
a. Survey Instrument:
In this study, quantitative research method approach was
used. To collect the data, a questionnaire form was designed to
achieve the research objectives. Based on extended literature
review, appropriate research constructs which had been
validated in prior studies were developed. These include
benefits, barriers, and motivation to use health information
systems. In addition, the questionnaire included a section of
general information about the respondents' demographics and
organizational variables which were considered as moderators
to the perceptions towards the health information applications.
The second section included 25 statements regarding the
benefits, barriers, and motivation of the health information
applications using five-point Likert-scale (1=Strongly
Disagree, 2=Disagree, 3=Neither agree nor disagree, 4=Agree,
5=Strongly agree). Thereafter, the questionnaire was validated
through evaluation by two faculty members of King Saud
University, and a pilot study. Cronbach's alpha values for the
three dimensions (benefits, barriers, and drivers) were strictly
above 0.74; meeting the recommended alpha threshold values
of at least 0.7 [23;24]. Therefore, all the three dimensions were
internally consistent.
b. Population and Sample
The study population consists of all healthcare providers
working at KAMC. The healthcare providers include
physicians, nurses, and clinical/paramedical personnel. The
population size is 7493. A sample size of 623 was drawn from
the population using convenience random sampling method.
Convenience sampling is a non-probability sampling technique
where subjects are selected because of their convenient
accessibility and accessibility to the researcher [25]. The
questionnaires with cover letters that explained the purpose of
the study were distributed during April/May 2011. Of 623
questionnaires distributed, 377 questionnaires were returned,
giving a response rate of 60.5 percent.
c. Analysis:
Descriptive statistics were used to analyze the demographic
and organization variables and the respondents' perceptions
towards benefits, barriers, and motivation to use health
information systems. One-sample t-test was conducted to
determine whether the mean score of each item of the three
dimensions (benefits, barriers, motives) is significantly higher
than a score 3; this being the mid-point on the Likert scale for
"Neither agree nor disagree" response to the item. Two-sample
t-test was used to test whether there are differences in
respondents' perceptions towards IT benefits, barriers to using
IT, and motives to using IT variables with respect to gender.
One-way analysis of variance (ANOVA) was used to test
whether there are difference in respondents' perceptions
towards IT benefits, barriers to using IT, and motives to using
IT variables with respect to education and occupation.
III. RESULTS:
a. Respondents' Characteristics:
Table I shows the profile of respondents by age, gender,
education, work experience, and occupation. The average
respondent's age was 36.2 years associated with a relatively
high standard deviation of 9.6 years. This shows the medical
workforce at KAMC is young. With respect to gender, the vast
majority of the respondents were female, 86.2 percent, while
the remaining 13.8 percent were males. The sex disproportion
is due to the fact that most of the sample were nurses, 55.2
percent; as nurses are usually females.
In terms of educational background, most of the
respondents, 79.3 percent, hold bachelor's degree, followed by
17.2 percent who had postgraduate degree and 3.4 percent had
high school education or less.
The work experience of the respondents ranged from one
year to twenty-six years. About 43 percent of the respondents
had less than five years of work experience; followed by 27.3
percent who had between five to nine years, 20.2 percent
between ten to fourteen years, 5.6 percent between fifteen to
nineteen years, and lastly 4.2 per cent had work experience of
more than twenty years. The average working experience was
7.1 years with a relatively high standard deviation of 5.5 years.
As for occupation, the table shows that about two thirds of
the sample were nurses, followed by 17.2 percent were
physicians and the same percentage were other medical staff.
The results show that the nurse-physician ratio is relatively
high; 3.8 nurses per physicians in the sample compared to 2.5
for the Ministry of Health [26].
TABLE I. PROFILE OF RESPONDENTS
Variables Frequency Percent
Age (Years):
Vol. 2, No. 10, 2011
16 | P a g e
Variables Frequency Percent
20 - 29 107 28.4
30 - 39 157 41.6
40 - 49 63 16.7
50 and above 50 13.2
(Mean =36.2, Std deviation= 9.6)
Gender:
Male 52 13.8
Female 325 86.2
Education:
High school or less 13 3.4
Bachelors degree 299 79.3
Postgraduate degree 65 17.2
Experience (years):
5 161 42.7
5 - 9 103 27.3
10 - 14 76 20.2
15 - 19 21 5.6
20 and above 16 4.2
(Mean =7.1, Std deviation=5.5)
Profession:
Physician 65 17.2
Nurse 247 65.5
Other 65 17.2
b. Literacy and Use of Information Technology
Table II presents the levels of IT applications' knowledge
and training and frequency of IT use. The results show that
about two-thirds of the respondents attended training courses in
information applications, while the remaining respondents,
34.5 percent, stated that they didn't attend any training course
in this field. With regards to information technology literacy,
82.8 percent of the respondents stated that they had good
knowledge and skills in the use of information applications,
whereas 17.2 percent had poor skills in the use of information
applications. However, the results show that most respondents
who had training in IT had also good IT applications skills. The
Chi-squared test confirmed there were significant relationship
between training and IT knowledge at 0.01 level of
significance. As can be seen from the table, 62.1 percent had
training in IT field had also good IT skills compared to 20.7
percent of the respondents who had no training and had good
IT knowledge. These results indicated that training has positive
effect on health providers' IT knowledge and skills.
Table II shows also that, the vast majority of the
respondents, 86.2 percent, reported that they always use
systems' applications, while 10.3 percent stated that they
sometimes systems' applications. The remaining 3.4 percent of
the respondents expressed that they rarely use systems'
applications in KAMC. However, the results show there was
significant relationship between frequency of systems'
application use and IT knowledge at 0.01 level of significance.
The results show that most respondents with good IT
applications skills use always KAMC computerized systems.
TABLE II. RELATIONSHIPS BETWEEN RESPONDENTS' IT KNOWLEDGE
WITH TRAINING AND FREQUENCY IT USE

Knowledge of IT
applications

Total Chi-square
Test
Good Poor
Training:

Had training in IT
234 13 247
62.1% 3.4% 65.5%
Did not have
training in IT
78 52 130
20.7% 13.8% 34.5%
All respondents
312 65 377
82.8% 17.2% 100%
2
= 75.7;
(sig.=0.000)
Frequency of IT
use:

Always

286 39 325
75.9% 10.3% 86.2%
sometimes

13 26 39
3.4% 6.9% 10.3%
Rarely

13 0 13
3.4% 0.0% 3.4%
All respondents
312 65 377
82.8% 17.2% 100%
2
= 72.0;
(sig.=0.000)
c. Perceptions of healthcare providers towards the benefits,
barriers, and motives to use information technology
applications in KAMC
Table III presents the perceptions of healthcare providers
towards benefits, barriers, and motives to use IT applications.
The high mean scores of the respondents' responses on benefits
of IT applications, ranged between 3.6 to 4.4, reveal that all the
respondents perceive that the information technology
applications in KAMC are valuable. Therefore, healthcare
providers believe that all information technology applications
are important and beneficial to both patients and KAMC
With regard to barriers to IT use, the mean scores of the
respondents' responses ranged between 2.6 to 3.4. This explains
that the respondents were split over the barriers to IT use in
KAMC. The results of the t-test show that the following
represent barriers to IT use in KAMC (items with p-values less
than or equal 0.05):
Vol. 2, No. 10, 2011
17 | P a g e

Insufficient number of computers

Time consuming

Low system performance

System being down frequently
The results of the t-test show that the following do not
represent barriers to IT use in KAMC (items with p-values
greater than 0.05):

Lack of training for the hospital staff

Lack of technical support

Incapability of the system

Lack of management support
As for drivers to IT use, the respondents' mean score on
items measuring the motives of IT use ranged from 3.58 to
3.89, implying the respondents agree with four motives shown
in the table. Therefore, it can concluded that healthcare
providers generally would be motivated to use IT applications
in KAMC by provision of new applications and training,
contribution in change hospital's work procedures, and
provision of technical support.
d. The effect of gender, occupation, and training on
respondents' perceptions towards IT benefits, barriers to
using IT, and motives to using IT variables:
1) Gender:
Two-sample t-test was used to test whether there are
differences in respondents' perceptions towards IT benefits,
barriers to using IT, and motives to using IT variables with
respect to gender. As for benefits of IT, Table IV shows that
there were significant differences in respondents' perceptions
on items 2, 4, 6, 7, 8, 9, and 13 with respect to gender at 0.05
significance level. The mean score of these items by gender
show that the female respondents rated items "provides speed
to accomplish work", "easier to find investigation results",
"facilitates coordination among departments", and "improves
quality of patients care" significantly higher than did male
respondents. Whereas, male respondents were more likely to
agree on items "prevent loss of patients` data", "helps in
preparing hospital reports ", and "improves decisions making
process" compared to female respondents.
TABLE III. RESPONDENTS' PERCEPTIONS TOWARDS BENEFITS, BARRIERS,
AND MOTIVES TO USE INFORMATION TECHNOLOGY APPLICATIONS IN KAMC
(N=377)
Item
M
e
a
n

s
c
o
r
e

S
t
d

d
e
v
i
a
t
i
o
n

t
-
v
a
l
u
e

P
-
v
a
l
u
e

Benefits of IT:

1. Easier to access patient records 4.4 0.61 43.3 0.000
2. Easier to find investigation
results 4.4 0.62 44.9 0.000
Item
M
e
a
n

s
c
o
r
e

S
t
d

d
e
v
i
a
t
i
o
n

t
-
v
a
l
u
e

P
-
v
a
l
u
e

3. Prevent loss of patients` data 4.3 0.71 36.7 0.000
4. Helps in preparing hospital
reports 4.3 0.81 30.8 0.000
5. Helps in managing patients 4.2 0.81 28.5 0.000
6. Provides speed to accomplish
work 4.1 0.77 28.3 0.000
7. Saving paper work 4.0 1.01 19.4 0.000
8. Facilitates coordination
among departments 4.0 0.99 18.9 0.000
9. Improves decisions making
process 4.0 0.86 21.8 0.000
10. Ensures patients` privacy 4.0 1.00 19.7 0.000
11. Reduces medical errors 3.9 0.70 24.1 0.000
12. Improves quality of patients
care 3.9 0.91 19.3 0.000
13. Decreases work load 3.6 1.27 8.5 0.000
Barriers to IT use:

1. Time consuming 3.4 1.18 5.9 0.000
2. Insufficient number of
computers
3.2 1.26 3.0 0.001
3. System being down frequently 3.1 1.12 2.5 0.007
4. Low system performance 3.1 1.08 1.6 0.053
5. Lack of training for the
hospital staff
2.9 1.07 -1.2 0.880
6. Lack of technical support 2.7 1.04 -5.9 1.000
7. Incapability of the system 2.6 0.87 -9.9 1.000
8. Lack of management support 2.6 0.98 -7.9 1.000
Motives to IT use:
1. Provide new / durable
applications 3.8 0.65 23.3 0.000
2. Provide training to staff 3.8 0.79 20.0 0.000
3. Change hospitals work
procedures 3.6 0.87 13.1 0.000
4. Provide technical support 3.9 0.62 28.1 0.000

The two-sample test's results show there were no significant
differences between male and female respondents in their
perceptions towards items 1, 3, 5, 10, 11, and12 at 0.05 level of
significance.
With respect to barriers to IT use, the results of two-sample
t-test show that there were significant differences in
perceptions of respondents on all items except for items 5, and
6 by gender. Male respondents indicated a higher agreement
with the first two barriers (insufficient number of computers
and time consuming) than did female respondents. While,
females were likely to agree on four barriers, low system
performance, system being down frequently, incapability of the
system, and lack of management support, than male
respondents. Furthermore, the results show there were no
significant differences in the perceptions of male and female
Vol. 2, No. 10, 2011
18 | P a g e
respondents towards "lack of training for the hospital staff" and
"lack of technical support" at 0.05 level of significance.
As for motives, the results show there were significant
differences between male and female respondents in their
attitudes towards the item which states "IT provides motives
new / durable applications" at 0.05 level of significance.
Female respondents indicated a higher agreement with the
statement compared with male respondents. Whereas, the
results show no other significant perceptions differences
between male and female respondents on the remaining items,
"provide training to staff", "change hospitals work procedures"
and "provide technical support", at 0.05 level of significance.
This means all health providers, regardless their gender, agreed
that these three items represent motives to IT applications in
KAMC
TABLE IV. RESULTS OF TWO-SAMPLE T-TEST OF IT BENEFITS, BARRIERS,
AND MOTIVES WITH RESPECT TO GENDER
Item
Mean Score
Two-sample t-
test
Male Female t-value
P-value
Benefits:

1. Easier to access patient
records
4.3 4.4 -1.0 0.300
2. Provides speed to
accomplish work
3.8 4.2 -3.4 0.001
3. Saving paper work 3.8 4.0 -1.8 0.068
4. Easier to find investigation
results
4.3 4.5 -2.1 0.041
5. Helps in managing patients 4.3 4.2 1.0 0.310
6. Facilitates coordination
among departments
3.3 4.1 -5.5 0.000
7. Prevent loss of patients`
data
4.5 4.3 2.0 0.049
8. Helps in preparing hospital
reports
4.5 4.2 2.1 0.032
9. Improves decisions making
process
4.3 3.9 2.6 0.011
10. Reduces medical errors 3.8 3.9 -0.7 0.474
11. Ensures patients` privacy 4.0 4.0 0.2 0.812
12. Decreases work load 3.8 3.5 1.3 0.196
13. Improves quality of
patients care
3.5 4.0 -3.3 0.001
Barriers:
c. Insufficient number of
computers
3.8 3.1 3.6 0.000
d. Time consuming 3.8 3.3 2.7 0.007
e. Low system performance 2.5 3.2 -4.5 0.000
f. System being down
frequently
2.3 3.3 -6.5 0.000
g. Lack of training for the
hospital staff
3.0 2.9 0.6 0.548
h. Lack of technical support 2.5 2.7 -1.4 0.170
i. Incapability of the
system
2.3 2.6 -2.7 0.006
j. Lack of management
support
2.0 2.7 -4.9 0.000
Motives:
1. Provide new / durable
applications
3.5 3.8 -3.2 0.002
2. Provide training to staff 3.8 3.8 -0.3 0.800
3. Change hospitals work
procedures
3.8 3.6 1.7 0.096
4. Provide technical
support
4.0 3.9 1.9 0.063
2) Occupation:
One-way Analysis of Variance (ANOVA) was used to
determine whether there was a significant mean difference in
the respondents perceptions on benefits of IT applications,
barriers to using IT applications, and motives to use IT
applications with respect to occupation. Table V demonstrates
the results of ANOVA test of IT benefits with occupation.
As for benefits, the results of the ANOVA tests showed that
there were significant differences between physicians, nurses
and other staff in their perceptions of all items measuring the
benefits of IT applications at 0.05 significance level. Nurses
had a higher positive perception than physicians and other staff
in the following items:

Provides speed to accomplish work.

Easier to find investigation results.

Helps in preparing hospital reports.

Improves decisions making process.

Decreases work load.
Whereas, other staff indicated higher agreement with the
following statements when compared to physicians and nurse:

Easier to access patient records

Saving paper work

Helps in managing patients

Facilitates coordination among departments

Prevent loss of patients` data

Reduces medical errors

Ensures patients` privacy

Improves quality of patients care
Interestingly, the physicians respondents indicated the
lowest agreement with all statements that measure the benefits
of IT applications in KAMC. These results are sensible since
all these benefits affect the performance of nurses and other
staff more than physicians. For example, from physicians
point of view IT does not decrease their workload, which is
why it was rated by them as low as 1.6 compared to 4 and 3.8
for nurses and other staff respectively.
As for barriers, the results of the ANOVA tests showed that
and other staff in their perceptions of all items measuring
barriers to use IT applications at 0.05 significance level; all p-
values were strictly less than 0.02. Physicians indicated higher
agreement with the following seven statements:
Vol. 2, No. 10, 2011
19 | P a g e


Time consuming






Whereas, the results show that the higher mean score of
respondents responses to item "System being down frequently"
was for other staff, followed by nurses and physicians. It is also
noted that nurse respondents were less likely to agree with
these stated barriers.
Table V presents the results of ANOVA test of motives IT
applications use with occupation as the factor.
Regarding motives, the ANOVA results showed that there
were significant differences between physicians, nurses and
other staff in their perceptions of all items measuring motives
to IT use at 0.01 significance level. Other staff respondents
indicated higher agreement with the statements "Provide
new/durable applications ", "Provide training to staff", and
"Provide technical support compared to physicians and nurses:
TABLE V. ONE-WAY ANALYSIS OF VARIANCE TEST OF IT BENEFITS
AND OCCUPATION
Item Mean score ANOVA
Physician Nurse Other Total F Sig.

Benefits:
Easier to access
patient records
4.2 4.3 4.6 4.4 7.3 0.001
Provides speed
to accomplish
work
3.6 4.2 4.2 4.1 17.3 0.000
Saving paper
work
2.6 4.3 4.4 4.0
121.5
0.000
Easier to find
investigation
results
4.2 4.5 4.5 4.4 4.7 0.010
Helps in
managing
patients
4.0 4.2 4.4 4.2 4.2 0.015
Facilitates
coordination
among
departments
3.0 4.1 4.2 4.0 48.1 0.000
Prevent loss of
patients` data
4.0 4.4 4.6 4.3 12.6 0.000
Helps in
preparing
hospital reports
4.0 4.4 4.0 4.3 10.5 0.000
Improves
decisions making
process
3.2 4.2 4.0 4.0 37.9 0.000
Reduces medical
errors
3.6 3.9 4.0 3.9 5.1 0.007
Ensures patients`
privacy
3.6 4.0 4.4 4.0 10.2 0.000
Decreases work
load
1.6 4.0 3.8 3.6 178.3 0.000
Item Mean score ANOVA
Physician Nurse Other Total F Sig.
Improves
quality of
patients care
3.0 4.0 4.6 3.9 72.4 0.000
Barriers:
Insufficient
number of
computers
3.6 3.1 3.2 3.2 4.3 0.015
Time
consuming
4.4 3.1 3.4 3.4 39.7 0.000
Low system
performance
3.8 2.9 3.2 3.1 21.6 0.000
System being
down
frequently
2.8 3.1 3.8 3.1 17.5 0.000
Lack of
training for
the hospital
staff
3.4 2.9 2.6 2.9 10.6 0.000
Lack of
technical
support
3.4 2.5 2.6 2.7 23.0 0.000
Incapability of
the system
3.0 2.4 2.6 2.6 12.6 0.000
Lack of
management
support
3.0 2.6 2.4 2.6 6.8 0.001
Motives:
Provide
new/durable
applications
3.0 3.9 4.0 3.8 81.2 0.000
Provide
training to
staff
2.8 4.0 4.2 3.8 109.2 0.000
Change
hospitals
work
procedures
3.2 3.7 3.4 3.6 13.6 0.000
Provide
technical
support
3.4 3.9 4.2 3.9 33.3 0.000
3) Training:
A two-sample t-test was performed to test whether there
were differences in respondents' perceptions towards IT
benefits, barriers and motives to using IT with respect to
training (Table VI). As for IT benefits, the results show that
there were significant differences (p-value < 0.05) in
perceptions of respondents who had training in IT and those
who had no training on all items except items "Ensures
patients' privacy" and "Improves quality of patients' care". It is
worth noting that the mean scores of the respondents who had
training on these items were higher than the mean scores of the
respondents who did not attend training in IT field. This shows
that the staff who attended training courses in IT perceive the
benefits of IT more than those who did not attend training
courses in this field. As shown in the table, the two-sample
test's results show that there were no significant differences in
the perceptions of staff who had training and those did not
attend training on items: "Ensures patients' privacy" and
"Improves quality of patients' care"
As for barriers, the results show there were significant
differences between respondents who attended training in IT
and those who did not attend training in their perceptions
Vol. 2, No. 10, 2011
20 | P a g e
towards barriers to using health information applications in
KAMC in the following items:

Time consuming



The mean scores of the respondents who did not attend
training on IT on these items were higher than the mean scores
of the respondents who had training except for the item "low
system performance". This indicates that the staff who attended
training courses in IT perceive less obstacles to IT use in
KAMC compared to staff who had no training in IT.
Conversely, the respondents who had training in IT perceive
that the system performance was low more than those who had
no training in IT. Moreover, the results show there were no
significant differences between respondents who had training
in IT and those who had no training in their perceptions
towards barriers to using health information applications in
KAMC in the following items:




With regards to drivers of IT use, the results show there
were significant differences between respondents who attended
training in IT and those who did not attend training in their
perceptions towards two items; "IT provides new/durable
applications" and "IT provides technical support at 0.05 level
of significance. The mean scores of the respondents who did
not attend training on IT on these items were higher than the
mean scores of the respondents who had training except for the
item "Provide new / durable applications". This reveal s that the
staff who attended training courses in IT perceive less
motivation to IT use in KAMC compared to those who had no
training in IT. On the contrary, the respondents who had
training in IT perceive that IT provides technical support more
than those who had no training in IT. Furthermore, there were
no significant differences between respondents who had
training in IT and those who had no training in their
perceptions towards two items; "IT provides training to staff"
and "IT changes hospitals work procedures".
TABLE VI. RESULTS OF TWO-SAMPLE T-TEST OF IT BENEFITS, BARRIERS,
AND MOTIVES WITH RESPECT TO ATTENDANCE OF TRAINING IN IT
Items
Mean Score Two-sample t-test
Attended
Not
attended
t-value P-value
Benefits:

Easier to access patient
records
4.5 4.0 8.0 0.000
Provides speed to accomplish
work
4.2 3.9 3.8 0.000
Saving paper work 4.2 3.7 3.9 0.000
Items
Mean Score Two-sample t-test
Attended
Not
attended
t-value P-value
Easier to find investigation
results
4.5 4.3 3.2 0.001
Helps in managing patients 4.3 3.9 4.5 0.000
Facilitates coordination
among departments
4.1 3.8 2.4 0.016
Prevent loss of patients` data 4.4 4.2 3.4 0.001
Helps in preparing hospital
reports
4.4 4.1 2.7 0.006
Improves decisions making
process
4.2 3.6 6.2 0.000
Reduces medical errors 4.0 3.6 5.9 0.000
Ensures patients` privacy 4.0 4.0 -0.3 0.747
Decreases work load 3.7 3.3 3.1 0.002
Improves quality of patients
care
4.0 3.8 1.5 0.134
Barriers:

Insufficient number of
computers
3.2 3.2 0.0 0.965
Time consuming 3.2 3.7 -4.2 0.000
Low system performance 3.2 2.9 2.6 0.010
System being down
frequently
3.1 3.2 -0.9 0.365
Lack of training for the
hospital staff
2.8 3.2 -3.6 0.000
Lack of technical support 2.5 3.0 -4.6 0.000
Incapability of the system 2.6 2.5 0.8 0.451
Lack of management support 2.6 2.7 -1.3 0.197
Motives:

Provide new / durable
applications
3.7 3.9 -2.9 0.004
Provide training to staff 3.9 3.7 1.9 0.057
Change hospitals work
procedures
3.6 3.6 -0.3 0.789
Provide technical support 4.0 3.7 3.7 0.000
IV. DISCUSSION
The results show that the majority of healthcare providers
use KAMC health information systems when the survey was
conducted. This result somewhat conflicts with Ananzy [19]
who found about 26 percent of healthcare providers use
electronic health records in six hospitals in Riyadh. Despite the
high HIT use, KAMC healthcare providers with good IT skills
used KAMC computerized systems more than those with poor
skills. This finding is consistent with that of Alam and Noor
[27, 28] who found significant effects of IT skills on adoption
of ICT.
The high mean scores of the respondents' responses on
benefits of HIT applications reveal that healthcare providers
perceive that the information technology applications in
KAMC are valuable and beneficial to both patients and
Vol. 2, No. 10, 2011
21 | P a g e
KAMC. This is consistent with findings of many researches
carried in USA which found that the healthcare providers
perceive the benefits of HIT in improving healthcare
[29,30].With regard to barriers, the healthcare providers were
split over the barriers to HIT use in KAMC. The healthcare
providers agreed that insufficient number of computers, time
consuming, low system performance, the system being down
frequently as barriers to HIT use in KAMC. Whereas, they
didn't perceive that lack of training for the hospital staff, lack of
technical support, incapability of the system, and lack of
management support were barriers to HIT use. These results
are somewhat consistent with Houser and Johnson [29]. As for
drivers, the results showed that healthcare providers generally
would be motivated to use IT applications in KAMC by
provision of new applications and training, contribution in
change hospital's work procedures, and provision of technical
support.
The results showed the perceptions of healthcare providers
on benefits, barriers and motives were influenced by gender.
However, the gender effect on perceptions of healthcare
providers is not consistent, as some items of the three
dimensions (benefits, barriers, and motives) were higher by
males and others were rated higher by females. However, there
were no significant differences in perceptions of some items
between male and female health providers. These results are to
some extent consistent with other research findings [31,32].
With respect to the effect occupation, the results show that
and other staff in their perceptions towards all items measuring
benefits, barriers, and motives. However, the effect of
occupation is also inconsistent; as some healthcare providers
had a higher positive perceptions than others. These results
conform with those of other research findings [33].
As regards the effect of training, the results show that
healthcare providers who attended training courses in IT
perceive the benefits of HIT more than those who did not
attend any training courses in this field. Similarly, the results
indicate that healthcare providers who attended training courses
in IT perceive less barriers to HIT use in KAMC compared to
those who had no training in IT. As for drivers of IT use, the
results show that the effect of training on motives to HIT use
were inconsistent as there were significant differences between
healthcare providers who attended training in IT and those who
did not attend training in their perceptions towards some items.
These results are consistent with previous research findings
which acknowledged the positive impact of training on IT
adoption [32,34].
The major research limitation of this study was the use of
convenience sample for data collection which might not
represented the target population accurately. Despite this
limitation and due to the lack of research in this area, the study
provides important information on the perceptions of
healthcare providers towards benefits, barriers and drivers of
health information technology in KAMC.
V. CONCLUSION , RECOMMENDATIONS AND FUTURE
SCOPE
The purpose of this study was to assess the perceptions of
benefits, barriers, and motives to use these applications. This
study also contributes in investigating the effects of gender,
occupation, and training on the perceptions of the healthcare
providers towards the health information applications.
Despite the perceived benefits and motives of health
information technology use, there were many barriers identified
by healthcare providers. The barriers include insufficient
number of computers, frequent system down, and that using
computerized systems is time consuming. Furthermore, there
were significant differences in the perceptions of healthcare
providers towards benefits, barriers, and motives to health
information technology with respect gender, occupation, and
training. Based on these results, the study recommends that
KAMC to provide easy access to health information
applications, continuous training to all healthcare providers on
health information technology, technical support services and
change hospital's work procedures. Further, the study also
recommends that KAMC administration to engage healthcare
providers in planning and promotion of health information
applications.
As a future scope, more research on the adoption of health
information technology applications can be carried out. The
scope can also be widened by considering the effect of
additional demographic and organizational variables on the
adoption HIT. Moreover, similar research can be carried in
other KAMC braches to trace geographic variations in HIT
adoption.
REFERENCES
[1] Kaye R, Kokia E, Shalev V, darD I, Chinitz D, Barriers and success
factors in health information technology: A practitioners
perspective, Journal of Management & Marketing in Healthcare.
2010; 3 (2): 163175.
[2] Jones, TM. National Infrastructure for eHealth: Considerations for
Decision Support. Studies in. Health Technology and Informatics,
2004; 100: 28-34
[3] Castells, M., Lupiez, F., Saig, F. and Snchez, J. E-Health and
Society: An Empirical Study of Catalonia Summary of the Final
Research Report, Catalan Internet Project, UOC and Generalitat de
Catalunya, Barcelona. 2007
[4] Hersh WR. Medical Informatics, Improving Healthcare through
Information. Journal of the American Medical Association. 2002;
288(16):1955-1958.
[5] Burton LC, Anderson GF, Kues IW. Using Health Records to Help
Coordinate Care. The Milbank Quarterly.2004; 82(3):457-481.
[6] Chaudhry B, Wang J, Wu S, Maglione M, Mojica W, Roth E, Morton
S, Shekelle P. Systematic review: Impact of health information
technology on quality, efficiency, and costs of medical care, Annals
of Internal Medicine. 2006; 144 (10):742752.
[7] McCullough J, Casey M, Moscovice I, Prasad S. The Effect of
Health Information Technology on Quality in US Hospitals. Health
Affairs. 2010; 29 (4):647-654.
[8] Hillestad R, Bigelow J, Bower Ay, Girosi F, Meili R, Scoville R and
Taylor R. Can Electronic Medical Record Systems Transform Health
Care? Potential Health Benefits, Savings, And Costs. Health Affairs,
2005;24(5):1103-1117.
[9] Institute of Medicine. To err is human: Building a safer health system.
ed. L. Kohn, J. Corrigan, and M. Donaldson. Washington, DC:
National Academy Press. 2000.
Vol. 2, No. 10, 2011
22 | P a g e
[10] Johnson KB, Ravich WJ, Cowan JA Jr. Brainstorming about next-
generation computer-based documentation. International Journal of
Medical Information. 2004; 73(9-10):665-674.
[11] Anderson, JG., Computer-based ambulatory information systems:
recent development, 2000
[12] Shekelle PG, Morton SC, Keeler EB. Costs and Benefits of Health
Information Technology. Evidence Report/Technology Assessment
No. 132. AHRQ Publication No.06-E006. Rockville, MD: Agency for
Healthcare Research and Quality. April 2006.
[13] Agarwal R, and Prasad J. The Antecedents and Consequents of User
Perceptions in Information Technology Adoption. Decision Support
Systems. 1998; 22:15-29.
[14] Venkatesh V, Morris M, Davis GB, Davis FD. User Acceptance of
Information Technology: Toward a Unified View. MIS Quarterly.
2003; 27(3):425-78.
[15] Morton ME. Use and Acceptance of an Electronic Health Record:
Factors Affecting Physician Attitudes. Unpublished dissertation,
Drexel University. 2008.
[16] Kim KK, Michelman J. An examination of factors for the strategic
use of information systems in the health care industry. MIS Qtly.
1990;14(2): 201-215.
[17] Chau P, Hu P. Examining a model of information technology
acceptance by individual professionals: An exploratory study. Journal
of Management Information Systems. 2001; 18:191-229.
[18] Devaraj S, Kohli R. Performance Impacts of Information
Technology: Is Actual Usage the Missing Link? Management
Science. 2003;49(3):273-289.
[19] Kohli R, Kettinger WJ. Informating the Clan: Controlling Physician
Costs and Outcomes. MIS Quarterly. 2004; 28(3): 363-394.
[20] Ministry of Economy and Planning. Ninth development plan; 2010-
2014. Ministry of Economy and Planning, Kingdom of Saudi Arabia.
2010.
[21] Alanazy, S. Factors Associated With Implementation of Electronic
Health Records in Saudi Arabia. Unpublished PhD dissertation
submitted to the University of Medicine and Dentistry of New
Jersey.2006.
[22] Harrison JP, Daly MA. Leveraging Health Information Technology
to Improve Patient Safety. Public Administration & Management.
2009; 13(3):218-237.
[23] De Vellis RE Scale development: Theory and application (2nd edn)
Thousand Oaks, California: Sage. 2003.
[24] Morgan GA, Gliner JA, Harmon RJ. Understanding and evaluating
research in applied clinical settings. Publisher: Lawrence Erlbaum
Associates; 2006.
[25] Roberts-Lombard M. Marketing Research A South African
Perspective. (3rd ed.). Cape Town: Oxford University Press. 2002.
[26] Ministry of Health. Health Statistical Year Book. Ministry of Health.
Kingdom of Saudi Arabia. 2009.
[27] Alam, SS, Noor, MKM. ICT Adoption in Small and Medium
Enterprises: an Empirical Evidence of Service Sectors in Malaysia.
International Journal of Business and Management. 2009; 4(2):112-
125.
[28] Hashim J. Information Communication Technology (ICT) Adoption
Among SME Owners in Malaysia. International Journal of Business
and Information; 2007; 2(2):221-240.
[29] Houser SH, Johnson L. Perceptions Regarding Electronic Health
Record Implementation among Health Information Management
Professionals in Alabama: A Statewide Survey and Analysis.
Perspectives in Health Information Management. 2008; 5; 6.
[30] Thakkar M, Davis DC. Risks, Barriers, and Benefits of EHR Systems:
A Comparative Study Based on Size of Hospital. Perspectives in
Health Information Management. 2006; 3(5):1-10.
[31] MacGregor R, Hyland P, Harvie C. (The Effect of Gender on
Perceived Benefits of and Drivers for ICT Adoption in Australian
Medical Practices. International Journal of E-Politics. 2011; 2(1):68-
85.
[32] Yu P, Li H, Gagnon M. Health IT acceptance factors in long-term
care facilities: A cross-sectional survey. International Journal of
Medical Informatics. 2009; 78:219-229.
[33] McDonald CJ, Overhage JM, Tierney WM, Dexter PR, Martin DK,
Suico JG, et al. The Regenstrief Medical Record System: a quarter
century experience. International Journal of Medical
Informatics1999;54:22553.
[34] Shachak A Fine S. The effect of training on biologists acceptance of
bioinformatics tools: A field experiment. Journal of American Society
for Information Science. 2008;59(5):719-730.

Vol. 2, No. 10, 2011

23 | P a g e
Route Maintenance Approach for Link Breakage
Predicttion in Mobile Ad Hoc Networks

Khalid Zahedi
Faculty of Computer Science and Information Systems
Universiti Teknologi Malaysia (UTM)
Johor, Malaysia
Abdul Samad Ismail
Faculty of Computer Science and Information Systems
Universiti Teknologi Malaysia (UTM)
Johor, Malaysia

Abstract Mobile Ad hoc Network (MANET) consists of a
group of mobile nodes that can communicate with each
other without the need of infrastructure. The movement of
nodes in MANET is random; therefore MANETs have a
dynamic topology. Because of this dynamic topology, the
link breakages in these networks are something common.
This problem causes high data loss and delay. In order to
decrease these problems, the idea of link breakage
prediction has appeared. In link breakage prediction, the
availability of a link is evaluated, and a warning is issued if
there is a possibility of soon link breakage. In this paper a
new approach of link breakage prediction in MANETs is
proposed. This approach has been implemented on the
well-known Dynamic Source Routing protocol (DSR). This
new mechanism was able to decrease the packet loss and
delay that occur in the original protocol.
Keywords- MANET; link breakage prediction; DSR.
I. INTRODUCTION
Mobile Ad hoc Network (MANET) consists of a group of
mobile nodes that can be communicated with each other
wirelessly without the need to any existed infrastructure.
MANETs in general are known with its dynamic topology.
The nodes are mobile and their movement is random.
MANETs dynamic topology makes link breakages a frequent
habit. This habit causes many problems such as data loss,
delay, and others which degrade the performance of the
MANETs protocols. In order to reduce the damage size of
this phenomenon, the idea of link breakage prediction has
appeared.
In link breakage prediction, a link breakage can be
predicted before its real occurring so route maintenance can
start before the occurring of the problem avoiding the
problems that come with a link breakage. In the link breakage
prediction, a node in an active route can predict if the link
between it and its previous hop will break soon. In this case it
can inform the source node about the problem and the source
node, if still needs the route, will be able to construct a new
route which avoids this soon to be broken link. It has been
found that this procedure has made a good improvement in the
performance of the mobile ad hoc networks protocols, but the
problem is that the focusing during constructing a new route
was only on excluding the link that was predicted to have a
link breakage. This mechanism may cause constructing a new
route with some or all bad links from the current used route
which are weak but did not predicted to be broken yet. These
links may break during or directly after the constructing of the
new route which will cause a high decrease in the packet
delivery ratio and a high increase in the packet loss and delay.
In order to improve the idea of link breakage prediction, this
paper has proposed a new approach for link breakage
prediction in MANETs. In this new approach, the source node
of an active route, after being informed about a link breakage
in its current used route, will construct a new route which
avoids the use of any link from the current used route. That
means excluding all the links in the current route, or in other
words, excluding the whole current used route not just the
soon to be broken link. So, the new constructed route will be
completely different from the current used one. This approach
is novel and it has been implemented on the well-known
reactive routing protocol Dynamic Source routing Protocol
(DSR).
This paper is organized in seven sections: Section I is an
introduction. Section II gives some examples of the works
that have been done in this area. Section III gives a
description about the Dynamic Source Routing protocol
(DSR). Section IV illustrates the proposed idea. Section V
discusses the simulation environment. Section VI detailed the
results that have obtained. Section VII concludes this paper,
and section VIII provides some future works.
II. LITERATURE REVIEW
Several researchers have investigated the area of link
breakage prediction in mobile ad hoc networks. In this
section, some examples of their works are discussed.
Ramesh et al. [1] have studied the problem of link
breakage prediction in the DSR routing protocol. Their idea is
that during the route discovery process, the source node builds
two routes which are the source route and another route can be
used as a backup. The backup route can be used if the primary
route (source route) was predicted to have a link breakage
soon.
Li et al. [2] have studied the link prediction in the AODV
routing protocol by establishing a signal intensity threshold
which is Pr-THRESHOLD. If the received signal intensity is
lower than the threshold, the upstream node will calculate the
distance between it and the sending node through the intensity
of the received packet signal, and estimate the relative velocity
between it and the sending node through the time difference of
Vol. 2, No. 10, 2011

24 | P a g e
the neighboring received data and the intensity of the packet
signal. Then, according to the relative position and the
relative velocity with the sending node, a node can estimate
when to send a RRER to the sending node to warning it about
a link failure. When the source node received this RRER
message, it will start its restored process searching its routing
table and find another route to the destination.
Qin & Kunz [3] have dealt with the problem of link failure
prediction by proposing an equation to calculate the exact time
that a link breakage can occur. They named their method the
link breakage prediction algorithm. In their idea, each node
maintains a table that contains the previous hop node address,
the value of the received packet signal power, and the time
which this data packet has been received. After receiving
three data packets, a node will calculate the link breakage time
and compare it with a fixed threshold. If the node predicted
that the link with its previous neighbor will have a link
breakage soon, it will send a warning message to the source
node of the active route to warn it about the link breakage
probability. If the source still needs the route it will perform a
route discovery process to establish a new route to the
destination. Their idea has been implemented using DSR
routing protocol.
Zhu [4] has studied the problem of link breakage
prediction by using the same equation that have been proposed
by Qin & Kunz [3] which is the link breakage prediction
algorithm, but she has implemented this algorithm using the
AODV and MAODV routing protocols
Choi et al. [5] has dealt with the problem of link breakage
prediction in vehicular ad hoc network. They proposed an
algorithm to predict a link breakage possibility using the value
of the RSSI (Received Signal Strength Indicator). Each
vehicle in the network periodically scans the received signals
from its neighbors and uses the collected value to calculate the
distance, the velocity, and the acceleration of its next hop
which it receives data packets from. By calculating these
three values, the node can predict if a link breakage will occur,
and can determine if the effected link can be maintained or a
new link is needed to be constructed. If the effected vehicle
found that a link breakage in the link with its next hop will
occur, it will use one of its neighbors which has the highest
value of RSSI with (that means the one which is the nearest to
it) to build a new link with before the previous link with its
other neighbor becomes broken.
Goff et al. [6] have studied the link breakage problem in
the DSR routing protocol. They defined a region they named
it the preemptive region, and they also defined a threshold
which they named it the preemptive threshold, they defined
this threshold as the signal power of the received packets at
the edge of the preemptive region. When a node enters the
preemptive region it will send a warning message to the source
node of the active route in order to inform it that a link
breakage will soon occur. So if the source is still interesting
with the route, it will generate a route discovery process to
establish a new route without that soon to be broken link.
Ouni et al. [7] studied the problem of link breakage
prediction in the DSR routing protocol and tried to propose a
solution by proposing a check model composed of two
modules. The first module includes performing different
simulations to have an idea about the nodes behavior and by
this allowing determining the suitable routes to use, while the
second model checks the path availability and the deadline
delay satisfaction. This check model was also used to predict
the validity periods of the selected path and the satisfaction of
the delay constrains.
Lu et al. [8] have worked on the DSR routing protocol and
proposed a mechanism for switching to a new route if the
current route is found to have a link breakage soon. Their
mechanism which is named DSR-link switch (DSR-LS) first
detects a link breakage between a nod and its next hop to the
source by measuring the power of the received packets. If a
link failure is detected to occur soon, the node, using this
mechanism, will send a link switch request (LSRE) in one hop
range to search appropriate nodes that act as relaying stations
or bridge nodes. This LSRE request will be sent by including
it in the RTS/CTS packets of the MAC layer during the
current communication. After finding a new strong links, the
current route will be shift to a more stable path.
III. DYNAMIC SOURCE ROUTING (DSR)
The Dynamic Source Routing (DSR) is a simple and
efficient routing protocol designed to be used specifically in
mobile ad hoc networks. Through using DSR, the network is
completely self-organizing and self-configuring. Network
nodes cooperate to forward packets for each other to allow
communication over multiple hops between the nodes that are
not located within the transmission range of each other. As
nodes in the mobile ad hoc network move about, join or leave
the network, and as wireless transmission conditions such as
types of interference change, all routing is automatically
determined and maintained by the DSR routing protocol.
The DSR routing protocol applies the idea of source
routing, this idea can be summarized by sending the whole
route from the source node to the destination node in each
transmitted IP packet, so the intermediate nodes will have to
only forward these packets without taking any routing
decision. In order to implement the idea of source routing,
DSR makes use of special header for carrying control
information which can be included in any IP packet. This
header is named DSR options header [9].
The DSR options header is a header existed in any sent IP
packet by a node implements DSR routing algorithm. This
header must immediately follow the IP header in the sent
packet. It consists of two fields, fixed length field and
variable length field. The fixed length field is a 4-octet
portion that has four fields (Next Header, F, Reserved,
Payload Length) while the variable length field is called the
options field, which has zero or more pieces of optional
information which are called DSR options. In DSR routing
protocol there are eight types of options, each one of them
must be included in a DSR options header in order to be
transmitted along the network.
DSR options header is located in an IP packet directly after
the IP header and before any other header in the packet. It can
contain one or more of the following options:
Vol. 2, No. 10, 2011

25 | P a g e
1) Route Request option.
2) Route Reply option.
3) Route Error option.
4) Acknowledgement request option.
5) Acknowledgement option.
6) DSR source route option.
7) Pad1 option.
8) PadN option.
The DSR protocol composes of two basic mechanisms
which work together to allow the discovery and maintenance
of the source routes in mobile ad hoc networks. These two
basic mechanisms are:
1) Route discovery
2) Route maintenance
Route discovery is the mechanism that is used by a source
node which wishes to send data packets to a destination node
which has no route to it in its route cache. Using this
mechanism the source node can obtain a source route to the
destination.
Route maintenance is the mechanism that is used by a
source node to detect a link breakage along its source route to
a destination node. Using this mechanism the source node can
know if it can still use the route or not. When the source node
indicates the existence of a broken link in the source route, it
can use another route or trigger a new route discovery process.
Route maintenance is used only with active routes.
Route discovery and route maintenance mechanisms each
operates entirely on demand. Unlike other protocols, DSR
does not require periodic packets of any kind at any level
within the network. For example, DSR does not use any
periodic routing advertisement and does not use neighbor
detection messages. This is a full on demand behavior.
It is possible that a link may not work equally well in both
directions because of antenna, or propagation patterns, or
sources interference. These types of links are called
unidirectional links. The routes that compose of such type of
links are called asymmetric routes or paths. DSR allows
unidirectional links to be used when necessary; this improves
the overall performance and the network connectivity.
DSR also supports the internetworking between different
types of wireless networks allowing a source route to be
composed of hops over a combination of any types of
networks available [10]. As an example, some nodes in the ad
hoc network may have only short-range radios, while other
nodes have both short-range and long-range radios; the
combination of these nodes together can be considered by
DSR as a single ad hoc network.
IV. THE PROPOSED IDEA
In this section a new approach for the link breakage
prediction in the mobile ad hoc networks will be introduced.
The idea is to construct a new route which is completely
different from the current used route by excluding all the links
exist in the current used one. So during the phase of
constructing the new route if another link or other links have
been predicted to be broken, there will be no need for trying to
avoid this link or these links, because from the beginning, the
new constructed route has excluded all the links in the
previous route. The approachs idea is as follows:
Each node along an active source route scans the received
data packets signals from its previous hop node. When a node
found that the Received Signal Strength Indicator (RSSI)
value of the received data packets from its previous hop is still
decreasing after three successive measurements, the node will
realize that the link between it and its previous hop will have a
link breakage soon. In this case it will generate a packet and
initialize a new option which will be named Soon Link
Breakage warning (SLBW). This option will be inserted in
the options field of the DSR options header of the packet.
Then, this packet which can be named SLBW message will be
unicasted to the source node of this active route to indicate to
it that a link breakage along this route will occur. The SLBW
option is similar to the RERR option of the DSR routing
protocol with some modifications, the error type in the SLBW
will be set to (4) in order to indicate the link breakage
probability. SLBW will include the source nodes address in
order to reach the source of the affected route in case more
than one route share some of the links of the affected route,
and will also include the addresses of both, the node that
predicted the link breakage and its previous hop nodes
address. By sending the addresses of the nodes at the end of
the soon to be broken link, the source node will be able to
determine which route will have a link breakage. When the
source node receives the SLBW message, if it still needs the
route, it will set the route that has a soon to be broken link
with the state of Route with a Breakage Prediction (RBP) in its
route cache. Then it will check its route cache to see if it has
another route to the destination. If it has one, it will make a
match between the intermediate node addresses of the cached
route and the node addresses in the current used route which
has the state (RBP). If there was no match, the source starts
sending data packets using this new source route. Otherwise,
it will trigger a route discovery process by broadcasting to its
neighbors a Modified Route Request (MRREQ) message. The
MRREQ message is an IP packet generated by the source node
which its DSR options header contains two options, the RREQ
option and the source route option.
In the source route option, the source node will append the
route with the (RBP) state. This step is made by the source
node in order to discover a new route which has no any
relationship with the current used route which has the state
(RBP), because the current route may have other weak links.
Each node receives this MRREQ message will check first if it
is the destination of this MRREQ. If it is the destination, it will
initialize a RREP option similar to the one in the original DSR
routing protocol. Else, it will check if it has received this
message before, so if the RREQ option in the received
MRREQ message has the same source address and REQUEST
ID of a previous received one, or if the receiving node found
its address appended in the RECORD of the received option, it
will discard this message. Otherwise, the node will check if
its address is appended in the source route option of the
MRREQ message. If it found its address appended, it will
discard the MRREQ message. Else, it will append its address
Vol. 2, No. 10, 2011

26 | P a g e
in the RECORD of the RREQ option in the MRREQ message
and rebroadcast the message to its neighbors.
In Fig. 1, in order to construct a route which has no any
relationship with the current used one, when node 1 receives
the MRREQ message it will make a match between its address
and the addresses in the source route option of the MRREQ
message. So when it found its address appended, it will
discard the message and not forward it any more. The same
situation will repeat with the other nodes of the route.
V. SIMULATION
In this section the parameters that have been manipulated,
the metrics that have been used for comparison, and the
environment that has been used to make the experiments will
be discussed in detail.
THE USED PARAMETERS
From our literature review, we found that most of the other
papers have used three parameters for making their
comparisons; these parameters are (number of nodes in the
network, simulation time, and pause time). In order to make
new and unique comparisons, we used in this paper three other
parameters which we found that no other paper in the link
prediction area has used before. These three parameters are:
1) Number of nodes per route.
2) Node mobility speed.
3) Node transmission range.
THE USED METRICS
In this paper three metrics have been used in order to make
the comparisons between the two protocols. These metrics
are:

Figure1. A clarification to the idea
1) Packet Delivery Ratio.
2) Number of dropped data packets.
3) Average End to End Delay.
The following is the definision of each metric:
Packet Delivery Ratio: It is the ratio between the
number of received data packets by the destination
and the number of generated data packets by the
source.
Number of dropped data packets: It is the number of
data packets that have failed to arrive successfully to
the destination.
Average End to End Delay: It is the time that is
taken by a packet in order to transfer from a source
node to a destination node.
THE USED ENVIRONMENT
As we mentioned, the simulations in this paper have been
carried out by varying three parameters. When any parameter
(of the three used parameters) is manipulated, all the other
parameters will be fixed. See Table I.
TABLE I. SIMULATION PARAMETERS

VI. RESULTS AND DISCUSSION
In this section, the achieved results will be discussed in
detail. In figure 2, we can see that the (Packet delivery ratio) is
decreasing for both protocols as the number of nodes per route
is increasing, but the decreasing in the case of (DSR modified)
is much less than the decreasing in the (DSR original). The
reason of decreasing in the PDR is that when the number of
nodes in the route increases this means that the number of
links in that route also increases, so the probability of link
breakages occurrence also increases. Also, we can notice that
the difference in PDR between the two protocols is big when
the number of nodes per route is low (as it is clear when there
is 10 nodes), but this difference is reduced gradually as the
number of nodes per route increases (as it is clear when there
is 100 nodes). The reason behind this is that the increase in
the number of nodes per route reduces the efficiency of the
new mechanism where link breakages will so frequently
occur.
In figure 3, we can see that the (Number of dropped data
packets) is increasing for both protocols as the number of
nodes per route is increasing, but the increasing in the case of
(DSR modified) is much less than the increasing in the (DSR
original). The reason of increasing in the number of dropped
data packets is that when the number of nodes in the route
increases this means that the number of links in that route also
increases, so the probability of link breakages occurrence also
increases. Also, we can notice that the difference in the
number of dropped data packets between the two protocols is
Vol. 2, No. 10, 2011

27 | P a g e
big when the number of nodes per route is low (as it is clear
when there is 10 nodes), but this difference is reduced
gradually as the number of nodes per route increases (as it is
clear when there is 100 nodes). The reason behind this is that
the increase in the number of nodes per route reduces the
efficiency of the new mechanism where link breakages will so
frequently occur.
In figure 4, we can see that the (Average End to End
Delay) is increasing for both protocols as the number of nodes
per route is increasing, but the increasing in the case of (DSR
modified) is much less than the increasing in the (DSR
original). The reason of increasing in the average end to end
delay is that when the number of nodes in the route increases
this means that the number of links in that route also increases,
so the probability of link breakages occurrence also increases.
Also, we can notice that the difference in the average end to
end delay between the two protocols is big when the number
of nodes per route is low (as it is clear when there is 10
nodes), but this difference is reduced gradually as the number
of nodes per route increases (as it is clear when there is 100
nodes). The reason behind this is that the increase in the
number of nodes per route reduces the efficiency of the new
mechanism where link breakages will so frequently occur.
In figure 5, we can see that the (Packet delivery ratio) is
decreasing for both protocols as the mobility speed of nodes is
increasing, but the decreasing in the case of (DSR modified) is
much less than the decreasing in the (DSR original). The
reason of decreasing in the PDR is that the increase in the
mobility speed of nodes forming a route means the increase in
the link breakages in the links between those nodes. Also, we
can notice that the difference in PDR between the two
protocols is big when the mobility speed of nodes is low (as it
is clear when it is 10 m/s), but this difference is reduced
gradually as the mobility speed increases (as it is clear when it
is 100 m/s).
packets) is increasing for both protocols as the mobility speed
of nodes is increasing, but the increasing in the case of (DSR
original). The reason of increasing in the number of dropped
data packets is that the increase in the mobility speed of nodes
forming a route means the increase in the link breakages in the
links between those nodes. Also, we can notice that the
difference in the number of dropped data packets between the
two protocols is big when the mobility speed of nodes is low
(as it is clear when it is 10 m/s), but this difference is reduced
is 100 m/s). The reason behind this is that the increase in the
mobility speed of nodes of the route reduces the efficiency of
the new mechanism where link breakages will so frequently
occur.
Delay) is increasing for both protocols as the mobility speed of
nodes is increasing, but the increasing in the case of (DSR
original). The reason of increasing in the average end to end
delay is that the increase in the mobility speed of nodes
forming a route means the increase in the link breakages in the
links between those nodes. Also, we can notice that the
difference in the average end to end delay between the two
protocols is big when the mobility speed of nodes is low (as it
is clear when it is 10 m/s), but this difference is reduced
is 100 m/s). The reason behind this is that the increase in the
mobility speed of nodes of the route reduces the efficiency of
the new mechanism where link breakages will so frequently
occur.
In figure 8, we can see that the (Packet delivery ratio) is
decreasing for both protocols as the transmission range of
nodes is decreasing, but the decreasing in the case of (DSR
modified) is much less than the decreasing in the (DSR
original). The reason behind the decreasing in the PDR is that
the decrease in the transmission range of nodes in a route
means that the links between those nodes will be weaker, so
the probability of breaking such links will be higher. Also, we
can notice that the difference in PDR between the two
protocols is big when the transmission range of nodes is high
(as it is clear when it is 750 m), but this difference is reduced
gradually as the transmission range decreases (as it is clear
when it is 10 m). The reason behind this is that the decrease in
the transmission range of nodes of the route reduces the
efficiency of the new mechanism where link breakages will so
frequently occur.
packets) is increasing for both protocols as the transmission
range of nodes is decreasing, but the increasing in the case of
original). The reason behind the increasing in the number of
dropped data packets is that the decrease in the transmission
range of nodes in a route means that the links between those
nodes will be weaker, so the probability of breaking such links
will be higher. Also, we can notice that the difference in the
number of dropped data packets between the two protocols is
big when the transmission range of nodes is high (as it is clear
when it is 750 m), but this difference is reduced gradually as
the transmission range decreases (as it is clear when it is 10
m). The reason behind this is that the decrease in the
transmission range of nodes of the route reduces the efficiency
of the new mechanism where link breakages will so frequently
occur.
Delay) is increasing for both protocols as the transmission
range of nodes is decreasing, but the increasing in the case of
original). The reason behind the increasing in the average end
to end delay is that the decrease in the transmission range of
nodes in a route means that the links between those nodes will
be weaker, so the probability of breaking such links will be
higher. Also, we can notice that the difference in the average
end to end delay between the two protocols is big when the
transmission range of nodes is high (as it is clear when it is
750 m), but this difference is reduced gradually as the
transmission range decreases (as it is clear when it is 10 m).
The reason behind this is that the decrease in the transmission
range of nodes of the route reduces the efficiency of the new
mechanism where link breakages will so frequently occur.
Vol. 2, No. 10, 2011

28 | P a g e
Figure 2 PDR and No. of nodes per route

Figure 3 No. of dropped packets and No. of nodes per route

Figure 4 Delay and No. of nodes per route

Figure 5 PDR and mobility speed

Figure 6 No. of dropped packets and mobility speed

Figure 7 Delay and mobility speed
Vol. 2, No. 10, 2011

29 | P a g e

Figure 8 PDR and transmission range

Figure 9 No. of dropped packets and transmission range

Figure 10 Delay and transmission range
VII. CONCLUSION
Many approaches have been proposed to deal with the
idea of link breakage prediction, but the problem is that all the
previous approaches were building a new route that avoids
using only the same soon to be broken link, but no one of
these approaches was able to build a new route which avoids
all the other links in the old route. In this paper, a new
approach for solving the problem of link breakages in
MANET has been proposed and implemented on the Dynamic
Source Routing (DSR) routing protocol. In this approach, the
Received Signal Strength Indicator (RSSI) value will be used
by a node along an active route to predict a link breakage in its
link with its next hop to the source node of this active route.
The node will warn the source node, and the source (if it still
needs the route) will discover a new route without using any
link from the current route which has a soon to be broken link.
The idea behind this is to reduce the probability of
constructing a route with bad links which can break during or
directly after the constructing of a new route. It has been
found that this approach was able to increase the packet
delivery ratio and decrease both the packet loss and the end to
end delay comparing to the DSR routing protocol. So, this
approach was able to improve the performance of the protocol.
VIII. FUTURE WORK
As a future work, this work can be extended by using
other metrics for making the comparisons between the original
and modified DSR routing protocols such as the terrain size,
packet size, packet sending rate, and others. Also, the traffic
mode can be changed from CBR to VBR and find the
difference. Another change can be made to the mobility
model. In this work the mobility model that has been used is
the random way point mobility model, so another research can
be done by using other mobility models such as the random
walk mobility model, or the random direction mobility model,
and see the difference.
REFERENCES
[1] Ramesh, V., Subbaiah, P., and Supriya, K. (2010). Modified DSR
(preemptive) to reduce link breakage and routing overhead for MANET
using proactive route maintenance (PRM). Global Journal of Computer
Science and Technology. Vol. 9. Issue 5, 124-129.
[2] Li, Q., Liu, c., and Jiang, H. (2008). The routing protocol AODV based
on link failure prediction. ICSP IEEE. 2008.
[3] Qin, L., and Kunz, T. (2002). Increasing packet delivery ratio in DSR by
link prediction. HICSS 03. IEEE. 2002. Hawaii.
[4] Zhu, Y. (2002). Proactive connection maintenance in AODV and
MAODV. Master of Science. Carleton University, Canada.
[5] Hoi, W., Nam, J., and Choi, S. (2008). Hop state prediction method
using distance differential of RSSI on VANET. NCM 2008. IEEE. 426-
431.
[6] Goff, T., Abu-Ghazaleh, N., Phatak, D., and Kahvecioglu, R. (2003).
Preemptive routing in ad hoc networks. Journal of Parallel and
Distributed Computing. 63 (2003), 123-140.
[7] Ouni, S., Bokri, J., and Kamoun, F. (2009). DSR based
routing algorithm with delay guarantee for ad hoc networks.
Journal of Networks. Vol. 4(5), 359-369.
[8] Lu, H., Zhang, J., and Luo, X. (2008). Link switch
mechanism based on DSR route protocol. ICINIS IEEE.
2008.
[9] Internet Engineering Task Force (IETF), Dynamic Source Routing
(DSR) protocol, 2007. http://www.ietf.org
Vol. 2, No. 10, 2011

30 | P a g e
[10] Sarkar, S. K., Basavaraju, T., Puttamadappa, C. (2008). Ad hoc mobile
wireless networks. (First ed.). New York: Auerbach Publications.
[11] Rajabhushanam, C., & Kathirvel, A. (2011). Survey of Wireless
MANET Application in Battlefield Operations. IJACSA - International
Journal of Advanced Computer Science and Applications, 2(1).
[12] Multicasting over Overlay Networks A Critical Review. IJACSA -
International Journal of Advanced Computer Science and Applications,
2(3), 54-61.
AUTHORS PROFILE
Khalid Zahedi is a Masters student in the faculty of Computer Science and
Information Systems in UTM. His research interests include: Mobile Ad hoc
Networks (MANETs) and Vehicular Ad hoc Networks (VANETs).
Abdul Samad Ismail is an associate professor in the faculty of Computer
Science and Information Systems in UTM. His research interests include:
Mobile Ad hoc Networks (MANETs), Vehicular Ad hoc Networks
(VANETs), Wireless Sensor Networks (WSNs), and Network Security.
Vol. 2, No. 10, 2011

31 | P a g e
Web Service Architecture for a Meta Search Engine

K.Srinivas
Associate Professor, Department of
IT, Geethanjali College of
Engineering & Technology,
Cheeryal(V), Keesara(M), Ranga
Reddy, Andhra Pradesh-501301,
India.
P.V.S. Srinivas
Professor & Head, Department of
CSE, Geethanjali College of
Engineering & Technology,
Cheeryal(V), Keesara(M), Ranga
Reddy, Andhra Pradesh-
501301,India.

A.Govardhan
Professor, Department of CSE,
School Of Information Technology,
Jawaharlal Nehru Technological
University Hyderabad, Kukapally,
Hyderabad,
Andhra Pradesh-500085, India.
AbstractWith the rapid advancements in Information
Technology, Information Retrieval on Internet is gaining its
importance day by day. Nowadays there are millions of Websites
and billions of homepages available on the Internet. Search
Engines are the essential tools for the purpose of retrieving the
required information from the Web. But the existing search
engines have many problems such as not having wide scope,
imbalance in accessing the sites etc. So, the effectiveness of a
search engine plays a vital role. Meta search engines are such
systems that can provide effective information by accessing
multiple existing search engines such as Dog Pile, Meta Crawler
etc, but most of them cannot successfully operate on
heterogeneous and fully dynamic web environment. In this paper
we propose a Web Service Architecture for Meta Search Engine
to cater the need of heterogeneous and dynamic web
environment. The objective of our proposal is to exploit most of
the features offered by Web Services through the implementation
of a Web Service Meta Search Engine.
Keywords-Meta Search Engine; Search engine; Web Services.
I. INTRODUCTION
A Meta Search Engine is a search tool that sends user
requests to several other search engines and/or databases and
aggregates the results into a single list or displays them
according to their source. Meta Search Engines enable users to
enter search criteria once and access several search engines
simultaneously [18]. More comprehensive search results can be
obtained by combining the results from several search engines.
This may save the user to use multiple search engines
separately.
This paper explores the web services and also the
Architecture of a Meta Search Engine. In section 2, we propose
the existing background and architecture of a Meta Search
Engine. Section 3 describes the architecture of a Web Service
framework. Section 4 describes about the Meta Search Engine
architecture, components and function of the each component.
In section 5, we propose the Web Service Meta Search Engine
architecture and Section 7 provides the advantages of Web
Service Metasearch Engines. Section 8 finally presents the
conclusions and the future work.
II. BACKGROUND
There are many methods for finding information, but one of
the leading ways is through search engines. Now a day,
everyone uses search engines for research, school, business,
shopping, or entertainment. So, the traffic on the Web is
growing exponentially [2]. To understand the working strategy
of a Search Engine, one should have an overview and clear
understanding of Information Retrieval System. Information
retrieval deals with the representation, storage, organization
and access to information items in order to give the user
desired information. A distinction between traditional
information retrieval and web information retrieval is that in
traditional or classic information retrieval, the process of
search is simple[6] and these document collections are stored in
physical form. An example would be looking for information
in books of a public library. Nevertheless, nowadays, most of
the documents are computerized that can be retrieved with the
help of a computer. Web information retrieval is on the other
hand, not like the traditional IR, where, searching is performed
within the globally largest collection of documents that are
linked, such as the well known search services on the internet
like Google or Yahoo [4].
Intense stress on user requirements recommended the
architecture of Meta Search Engine. A few Meta Search
Engines has been already proposed that provides quick
response with re ranked results after extracting user preference.
It uses Nave Bayesian Classifier for re ranking. An enhanced
version of open source Helios Meta Search Engine takes input
keywords along with specified context or language and gives
refined results as per users need [8]. All the proposed solutions
refine search-results up to some extent but they have a serious
drawback, which is that the user profile is not stationary. The
idea behind metasearch is to use multiple helper search
engines to do the search, then to combine the results from these
engines. Engines that use metasearch include Metacrawler,
SavvySearch, MSN Search and Altavista, among others [11].
III. WEB SERVICES
Web service protocol is designed for providing service via
web. Web Services are emerging as the fundamental building
blocks for creating distributed, integrated and interoperable
solutions across the Internet. They represent a new paradigm in
distributed computing that allow applications to be created
from multiple Web Services dispersed across the web
originating from various sources regardless of where they
reside or how they were implemented [19]. Unlike services in
Vol. 2, No. 10, 2011

32 | P a g e
general, Web services are based on specifications for data
transfer, method invocation and publishing. This is often
misunderstood and when a Web service is mentioned it
sometimes refers to a general service provided on the Web, like
the weather forecast on a Web page for example. The weather
forecast is a service and provides its functionality for a variety
of users but unless it comprises an interface to communicate
with other applications via SOAP [16]. Web services can be
seen as software components with an interface to communicate
with other software components. They have a certain
functionality that is available through a special kind of Remote
Procedure Call. In fact they even evolved from traditional
Remote Procedure Calls.There are various aspects of Web
services. Messaging, discovery, portals, roles, and
coordination. The format of the messages exchanged between a
client and a Web service is specified by a standard called
SOAP. The Simple Object Access Protocol (SOAP) is defined
by the W3C as a lightweight protocol for exchange of
information in a decentralized, distributed environment. SOAP
is an XML based protocol that consists of three parts: an
envelope that defines a framework for describing what is in a
message and how to process it, a set of encoding rules for
expressing instances of application-defined data types, and a
convention for representing remote procedure calls and
responses [12],[16].

Web service messages are sent across the network in an
XML format defined by the W3C SOAP specification. In most
Web services, there are two types of SOAP messages: requests
and responses. When a SOAP request is received, the Web
service performs an action based on the request and returns a
SOAP response. In many implementations, SOAP requests are
similar to function calls with SOAP responses returning the
results of the function call [12]. With SOAP, a communication
between Web services is possible and structured and each
participant knows how to send or receive the corresponding
SOAP Message. The final step to complete the communication
architecture of Web services is to define how to access a
service once it is implemented.

Figure 1.
Web services are the core any type of service, as it holds
and connects everything together as shown in Figure 1.
IV. META SEARCH ENGINE
Meta Search Engines can be classified into two types. a)
General purpose metasearch engine and b) Special purpose
Meta Search Engines. The former aims to search the entire
Web, while the latter focuses on searching information in a
particular domain (e.g., news, jobs).

Major Search Engine Approach: This approach uses a
small number of popular major search engines to build a
metasearch engine. Thus, to build a general-purpose
metasearch engine using this approach, we can use a
small number of major search engines such as Google,
Yahoo!, Bing (MSN) and Ask. Similarly, to build a
special purpose Meta Search Engine for a given domain,
we can use a small number of major search engines in that
domain.

Large scale Metasearch engine approach: In this approach,
a large number of mostly small search engines are used to
build a Metasearch engine. For example, to build a
general-purpose metasearch engine using this approach,
we can perceivably utilize all documents driven search
engines on the Web. Such a metasearch engine will have
millions of component search engines. Similarly to build a
special purpose metasearch engine for a given domain
with this approach, we can connect to all the search
engines in that domain. For instance, for the news domain,
tens of thousands of newspaper and news-site search
engines can be used.

Each of the above two approaches has its advantages and
disadvantages. An obvious advantage of the major search
engine approach is that such a metasearch engine is much
easier to build compared to the large-scale metasearch engine
approach because the former only requires the metasearch
engine to interact with a small number of search engines.
Almost all currently popular metasearch engines, such as
Dogpile, Mamma and MetaCrowler, are built using the major
search engine approach, and most of them use only a handful
of major search engine. One example of a large-scale special-
purpose metasearch engine is AllInOneNews, which uses about
1,800 news search engines from about 200 countries/regions.
In general, more advanced technologies are required to build
large-scale metasearch engines. As these technologies become
more mature, more large-scale metasearch engines are likely to
be built.
V. PROPOSED ARCHITECTURE
The proposed architecture takes both approaches into
consideration for designing web service metasearch engine
system. Significant software components included in this
architecture search engine selector, search engine connectors,
result extractors and result merger.
Vol. 2, No. 10, 2011

33 | P a g e

Figure 2 Web Service Meta Search Engine
Architecture
A. Search Engine Selector:
If the number of component search engines in a metasearch
engine is very small, say less than 10, it might be reasonable to
send each user query submitted to the metasearch engine to all
the component search engines. In this case, the search engine
selector is probably not needed. However, if the number of
component search engines is large, as in the large scale Meta
Search Engine scenario, then sending each query to all
component search engines will be an inefficient strategy
because most component search engines will be useless with
respect to any particular query. For example, suppose a user
wants to find 50 best matching results for his/her query from a
metasearch engine with 1,000 component search engines. Since
the 50 best results will be contained in no more than 50
component search engines, it is clear that at least 950
component search engines are useless for this particular query.
Passing a query to useless search engines may cause serious
problems for efficiency. Generally, sending a query to useless
search engines will cause waste of resources to the metasearch
engine server, each of the involved search engine servers and
the Internet. Specifically, dispatching a query, including needed
query reformatting, to a useless search engine and handling the
returned results, including receiving the returned response
pages, extracting the result records from these pages, and
determining whether they should be included in the final
merged result list and where they should be ranked in the
merged result list if they are to be included, waste the resources
of the metasearch engine server; receiving the query from the
metasearch engine, evaluating the query, and returning the
results back to the metasearch engine waste the resources of
each search engine whose results end up useless; and finally
transmitting a query from the metasearch engine to useless
search engines and transmitting useless retrieved results from
these search engines to the metasearch engine waste the
network resources of the Internet. Therefore, it is important to
send each user query to only potentially useful search engines
for processing.

The problem of identifying potentially useful component
search engines to invoke for a given query is the search engine
selection problem, which is sometimes also referred to as
database selection problem, server selection problem, or query
routing problem. Obviously, for metasearch engines with more
component search engines and/or more diverse component
search engines, having an effective search engine selector is
more important.
B. Search Engine Connectors:
After a component search engine has been selected to
participate in the processing of a user query, the search engine
connector established a connection with the server of the search
engine and passes the query to it. Different search engines
usually have different connection parameters. As a result, a
separate connector is created for each search engine. In general,
the connector for a search engine S needs to know the HTTP
(Hyper Text Transfer Protocol) connection parameters
supported by S. There are three basic parameters, (a) the name
and location of the search engine server, (b) the HTTP request
method (usually it is either GET or POST) supported by S, and
(c) the name of the string variable that is used to hold the actual
query string.
When implementing metasearch engines with a small
number of component search engines, experienced developers
can manually write the connector for each search engine.
However, for large-scale metasearch engines, this can be very
time-consuming and expensive. Thus, it is important to develop
the capability of generating connectors automatically.
An intelligent metasearch engine may modify a query it
receives from a user before passing it to the search engine
connector if such a modification can potentially improve the
search effectiveness. For example, a query expansion technique
may be used by the metasearch engine to add terms that are
related to the original user query to improve the chance for
retrieving more relevant documents.
C. Web Service:
A Meta search engine is a search engine that collects results
from other search engine. Web service offers such functionality
and then presents a summary of that information as the results
of a search. Most search engines available on the Web provide
only a browser based interface; however, because Web services
Vol. 2, No. 10, 2011

34 | P a g e
start to be successful, some of those search engines offer also
an access to their information through Web services. Two types
of search engines are observed, one that acts like a wrapper for
the HTML pages returned by the search engine and other one is
build upon the Web service offered by the search engine but
this difference is visible only when looking at the internal
processing of the service. It is difficult to distinguish them from
the outside as they implement the same interface.
Web service are built for, any process that can be integrated
into external systems through valid XML documents over
Internet protocols. This definition outlines the general idea of
Web services. Web services can be seen as software
components with an interface to communicate with other
software components. They have a certain functionality that is
available through a special kind of Remote Procedure Call.
SOAP, the Simple Object Access Protocol [16] was
developed to enable a communication between Web services. It
was designed as a lightweight protocol for exchange of
is an extensible, text-based framework for enabling
communication between diverse parties that have no prior
knowledge of each other. This is the requirement a transport
protocol for Web services has to fulfill. SOAP species a
mechanism to perform remote procedure calls and therefore
removes the requirement that two systems must run on the
same platform or be written in the same programming
language.
D. Result Extractors:
After a component search engine processes a query, the
search engine will return one or more response pages. A typical
response page contains multiple (usually 10) search result
records, each of which corresponds to a retrieved Web page,
and it typically contains the URL and the title of the page, a
short summary (called snippet) of the page content, and some
other pieces of information such as the page size. Fig 2 shows
the upper portion of a response page from the Google search
engine. Response pages are dynamically generated HTML
documents, and they often also contain content unrelated to the
user query such as advertisements (sponsored links) and
information about the host Web site.

A program (i.e., result extractor) is needed to extract the
correct search result records from different component search
engines can be merged into a single ranked list. This program is
sometimes called an extraction wrapper. Since different search
engines often format their results differently, a separate result
extractor is usually needed for each component search engine.
Although experienced programmers can write the extractors
manually, for large-scale metasearch engines, it is desirable to
develop techniques that can generate the extractors
automatically.
E. Result Merger:
After the results from the selected component search
engines are returned to the metasearch engine, the result merger
combines the results into s single ranked list. The ranked list of
search result records is then presented to the user, possible 10
records on each page at a time, just like more search engines
do. Many factors may influence how result merging will be
performed and what the outcome will look like. The
information that could be utilized includes the local rank of a
result record from a component search engine, the title and the
snippet of a result record, the full document of each result, the
publication time of each retrieved document, the potential
relevance of the search engine with respect to the query from
where a result is retrieved, and more. A good result merger
should rank all returned results in descending order of their
desirability.
The existing architecture has many disadvantages in search
engine selection, search engine connection and result
extractors. We proposed a robust metasearch engine
architecture using web services for heterogeneous and dynamic
environment.
VI. IMPLEMENTATION
WSMSE is a web based Meta Search Engine that is
developed using Java Web Service, Servlet and JSP. This
search engine is developed based on the assumption that an
average web user makes searches based on imprecise query
keywords or sentences, which in turn leads to unnecessary or
inaccurate results. In this work, we demonstrate that with a
simple tweak or manipulation of existing search engine
functions, which can helps users to get better search results.

To build and deploy Web service Java Web Service
Developer Pack (WSDP) will be used. Based on the Java 2
SDK, this toolset adds new API including XML Messaging
(JAXM), XML Processing (JAXP), XML Registries (JAXR),
XML-based RPC (JAX-RPC) and the SOAP with Attachments
(SAAJ). The main advantage Java WSDP is it supported
heterogeneous platform and have dynamic behavior.
Every Web service is deployed with a Web Service
Description Language (WSDL) file that acts like an instruction
manual; in case someone wants to use a particular Web
Service, he simply has to look at that WSDL file to learn how
to communicate with the corresponding service and use it.
Amongst all advantages, the most important one for the Web
services is that XML is not platform dependent and it allows
easy data processing and exchange between different
applications.
HTTP is the most popular transfer protocol and its
supported on almost all platforms. By implementing this
standard, combined with XML, Web services remove almost
all frontiers between platforms.
The Simple Object Access Protocol (SOAP) exchange of
is an XML based protocol that consists of three parts: an
envelope that defines a framework for describing what is in a
message and how to process it, a set of encoding rules for
expressing instances of application-defined data types, and a
convention for representing remote procedure calls and
responses.
Vol. 2, No. 10, 2011

35 | P a g e

Figure 3 SOAP request to the WSMSE, invoking the Get Available Search
Engine Web Service
Web Service Description Language (WSDL) defines an
XML grammar for describing network services as a set of
endpoints that accept messages containing either document-
oriented or procedure-oriented information. The endpoint is
defined by a network protocol and a message format, however,
the extensible characteristic of WSDL allow the messages and
endpoints being described regardless of what message formats
or network protocols are being used to communicate. In other
words, a WSDL file is an XML document that describes a set
of SOAP messages and how the messages are exchanged.
Universal Discovery Description and Integration (UDDI) is
the yellow pages of Web services. A UDDI directory entry is
an XML file that describes a business and the services it offers.
Both can be categorized and have keys so one can find a
provider or a service by different ways. Web services are also
defined through a document called a Type Model (tModel) that
describes their interface. A tModel is simply a WSDL file
without the <service> tag; it contains information to generate
the different proxy classes or SOAP messages to communicate
with the Web service but it does not specify the access point of
the service.
VII. ADVANTAGES OF WEB SERVICE META SEARCH ENGINE
We attempt to provide a comprehensive analysis of the
potential advantages of metasearch engines over search
engines. We will also focus on the comparison of metasearch
engine and search engine.
A. Increased Search Coverage:
Metasearch engine can search any document that is indexed
by at least one of the search engines. Hence, the search
coverage of a metasearch engine is the union of those search
engines. Metasearch engine with multiple major search engines
as components will have larger coverage than any single
component search engine. Different search engines often
employ different document representation and result ranking
techniques, and as a result, they often return different sets of
top results for the same user query. Thus, by retrieving from
multiple major search engines, a metasearch engine is likely to
return more unique high quality results for each user query.
B. Better Content Quality:
The quality of the content of a search engine can be
measured by the quality of the documents indexed by the
search engine. The quality of a document can in turn be
measured in a number of ways such as the richness and the
reliability of the contents. General-purpose metasearch engines
implemented using the large-scale metasearch engine approach
has a better chance to retrieve more up-to-date information than
major search engines and metasearch engines that are built with
major search engines.
C. Good Potential for Better Retrieval Effectiveness:
More unique results are likely to be obtained, even among
those highly ranked ones, due to the fact that different major
search engines have different coverage and different document
ranking algorithms. The result-merging component of the
metasearch engine can produce better results by taking
advantage of the fact that the document collection of major
search engines has significant overlaps. This means that many
shared documents have the chance to be ranked by different
search engines for any given query. If the same document if
retrieved by multiple search engines, then the likelihood that
the document is relevant to the query increases significantly
because there is more evidence to support its relevance. In
general, if a document if retrieved by more search engines, the
document is more likely to be relevant.
D. Better Utilization of Resources:
Metasearch engine use component search engines to perform
basic search. This allows them to utilize the storage and
computing resources of these search engines.
VIII. CONCLUSIONS AND FUTURE WORK
The goal of this proposal is to shed some light on the reasons
that make building metasearch engines, especially large scale
metasearch engines, difficult and challenging. Modern search
engines providing interfaces that allow external applications to
issue Web search queries that are actually processed using their
large scale computing infrastructure. This paper proposes a
robust Meta search engine, which can communicate to
heterogeneous platform using advanced web service
techniques. As much progress has been made in advanced
metasearch engine technology, several challenges need to be
addressed before truly large scale Meta Search Engine can be
effectively built and manage.
REFERENCES
[1] Lee Underwood, A Brief History of Search
Engines;www.webreference.com/authoring/search_history/.
[2] Ritu Khare, Yuan An and II-Yeol Song on Understanding Deep Web
Search Interfaces: A Survey. SIGMOD Record, March 2010 (Vol. 39,
No 1).
[3] DogPile metasearch engine. http://www.dogpile.com.
<?xml version="1.0" encoding="utf-8"?>
<soap:Envelope
xmlns:soap="http://schemas.xmlsoap.org/soap/envelop
e/"
xmlns:soapenc="http://schemas.xmlsoap.org/soap/enc
oding/" xmlns:tns="http://adv/WSMSE"
xmlns:types="http://adv/WSMSE/encodedTypes"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-
instance"
xmlns:xsd="http://www.w3.org/2001/XMLSchema">
<soap:Body
soap:encodingStyle="http://schemas.xmlsoap.org/soap
/encoding/">
<tns:GetAvailableSEWS />
</soap:Body>
</soap:Envelope>
Vol. 2, No. 10, 2011

36 | P a g e
[4] Carol L. Barry. The Identification of User Criteria of Relevance and
Document Characteristics: Beyond the Topical Approach to Information
Retrieval. PhDthesis, Syracuse, 1993.
[5] Ana B. Benitez, Mandis Beigi, and Shih-Fu Chang. Using
relevance feedback in content & based image metasearch. IEEE
Internet Computing,2(4):58-69, 1998.
[6] Eric J. Glover, Steva Lawrance, William P. Birmingham and C. Lee
Giles on Architecture of the Metasearch Engine that Supports User
Information Needs, ACM.
[7] Tieli Sun Wei Zhao and Zhiyan Zhao An Architecture of Pattern-
Oriented Distributed Meta-Search Engine ACM.
[8] Amir Hossein Keyhanipoor, Maryam Piroozmand, Behzad Moshiri and
Caro Lucas A Multi layer/Multi-Agent Architecture for Meta-Search
Engine, AIML 05 Conference, 19-21 December 2005, CICC, Cairo,
Egypt.
[9] Jae Hyun Lim, Young-Chan Kim, Hyonwoo Seung, Jun Hwang ,
Heung-Nam Kim, Query Expansion for Intelligent Information
Retrieval on Internet, International Conference on Parallel and
Distributed Systems, 1997.
[10] Eric J. Glover, Steve Lawrence, William P. Birmingham, C. Lee Giles;
Architecture of a Meta-Search Engine that Supports User Information
Needs, Information and knowledge management CIKM '99, ACM
Press,Pages: 210 - 216, 1999.
[11] Sergey Brin and Lawrence Page, The Anatomy of a Large-Scale
Hypertextual Web Search Engine, World-Wide Web Conference, 1998.
[12] Candadai, A Dynamic Implementation Framework for SOA-based
Applications, Web Logic Developers Journal: WLDJ,
September/October 2004.
[13] UDDI specifications http://www.uddi.org.
[14] SOAP specifications and RFCs - http://www.w3.org/TR/soap/.
[15] WSDL Specifications and RFCs - http://www.w3.org/TR/wsdl.
[16] T. Clements. Overview of SOAP. Java Developers Forum,
http://java.sun.com/developer/technicalArticles/ xml/webservices/.
[17] Ethan Cerami. Web Services Essentials. O'Reilly, February 2002.
[18] K.Srinivas, P.V.S.Srinivas and A.Govardhan. A Survey on the
Performance Evaluation of Various Meta Search Engines IJCSI
Volume 8, Issue 3,No. 2,May 2011 Pages 359-364.
[19] Nath, R. (2011). An Authorization Mechanism for Access Control of
Resources in the Web Services Paradigm. IJACSA - International
Journal of Advanced Computer Science and Applications, 2(6), 36-42.

Vol. 2, No.10, 2011

37 | P a g e
A Statistical Approach For Latin Handwritten Digit
Recognition

Ihab Zaqout
Dept. of Information Technology
Faculty of Engineering & Information technology
Al-Azhar University Gaza
Gaza Strip, Palestine

Abstract A simple method based on some statistical
measurements for Latin handwritten digit recognition is
proposed in this paper. Firstly, a preprocess step is started with
thresholding the gray-scale digit image into a binary image, and
then noise removal, spurring and thinning are performed.
Secondly, by reducing the search space, the region-of-interest
(ROI) is cropped from the preprocessed image, then a freeman
chain code template is applied and five feature sets are extracted
from each digit image. Counting the number of termination
points, their coordinates with relation to the center of the ROI,
Euclidian distances, orientations in terms of angles, and other
statistical properties such as minor-to-major axis length ratio,
area and others. Finally, six categories are created based on the
relation between number of termination points and possible
digits. The present method is applied and tested on training set
(60,000 images) and test set (10,000 images) of MNIST
handwritten digit database. Our experiments report a correct
classification of 92.9041% for the testing set and 95.0953% for
the training set.
Keywords- Digit recognition; freeman chain coding; feature
extraction; classification.
I. INTRODUCTION
The significant task of handwritten digit recognition has
great importance in the recognition of postcodes sort mail, bank
check amounts and so on. Since three decades, there is no
single classifier performs the best for all pattern classification
problems consistently. There are different challenges faced
while attempting to solve this problem. The handwritten digits
are not always of the same thickness, size, orientation or
position relative to margins.
There are several approaches for handwritten digit
recognition problem have been reported in the literature in the
past. They include SVM (support vector machine) [1, 2, 3], NN
(neural network) [4, 5, 6, 7], deformable template matching [8,
9], hybrid method [10, 11, 12, 13] and others. In SVMs for
digit classification problems, the training of a large data set is
still a bottle-neck and is comparatively slow. NNs have been
widely used to solve complex classification problems.
However, A single NN often exhibits with the over fitting
behavior which results in a weak generalization performance
when trained on a limited set of training data. A better
deformation algorithms and proper selection of representative
prototypes along with its computational requirements are
required for deformable template matching method. Hybrid
method has been widely used in pattern recognition
applications. It combines two or more of the above mentioned
methods or others to overcome their individual weakness and
to preserve their individual advantages. But it is still an open
problem to obtain a superior hybrid method.
A comprehensive benchmark of handwritten digit
recognition with several state-of-the-art approaches, datasets,
and feature representations had been reported by [14]. Several
classifiers and feature vectors are evaluated on MNIST
handwritten digit database [15].
In this paper, our research is focused on an accurate and
feasible method applied and tested on training set (60,000
images) and test set (10,000 images) of MNIST handwritten
digit database. We agree with others that the preprocess stage is
a crucial and it reflects the accuracy of the classification
process. Firstly, a preprocess step is started with thresholding
the gray-scale digit image into a binary image, and then noise
removal, spurring and thinning are performed. Secondly, by
reducing the search space, the region-of-interest (ROI) is
cropped from the preprocessed image, then a freeman chain
code template is applied and five feature sets are extracted from
each digit image. Counting the number of termination points,
their coordinates with relation to the center of the ROI,
area and others.
Finally, six categories are created based on the relation
between number of termination points and possible digits. The
advantage of method is that it does not require training, which
can save a lot of training time. Experimental results on MNIST
database will be reported in the paper to support the feasibility
of our method. The remainder of this paper is organized as
follows. The system overview is presented in Sec. 2. The
feature extraction is proposed in Sec. 3. In Sec. 4, the
classification is discussed. Experimental results are shown in
Sec. 5 to demonstrate the reliability of our method. Finally our
conclusion and future work is given in Sec 6.
II. SYSYEM OVERVIEW
In this section, the system overview is introduced. The
recognition system includes three dependent stages: the
Vol. 2, No.10, 2011

38 | P a g e
preprocess, the feature extraction and classification as shown in
Figure 1.

Figure 1. The System Overview
In this paper, a MNIST digit database is used as a dataset to
the proposed classification processes. A preprocessed step is
started with thresholding the gray-scale digit image into a
binary image, and then noise removal, spurring and thinning
[16] are performed. During the process of image thinning, we
introduce Freeman chain code tracking [17] as shown in Figure
2. To reduce the search space, the region-of-interest (ROI) is
cropped from the preprocessed image, then a freeman chain
code template is applied and five feature sets are extracted from
each digit image. Counting the number of termination points,
their coordinates with relation to the center of the ROI,
area and others. Six categories are created based on the relation
between number of termination points and possible digits.

Figure 2. Freeman chain code and 3x3 template
III. FEATURE EXTRACTION
A fast recognition system has to consider the processing
time and its complexity is primarily determined by the feature
extraction. In this section, the feature extraction is performed
on the thinned image. Five feature sets are extracted from each
ROI digit image:
Number of termination points, xy-coordinates and
orientations,
Minimum-to-maximum distance ratio (Euclidean
distance between the xy-coordinate of the center of
the ROI and xy-coordinates of termination points),
minor-to-major axis length ratio and their spatial
distribution compared to the center of the ROI,
Sum of white pixels after partitioning the thinned
digit image into horizontal blocks for 0, 2, 3, 5, 6 and
8,
Other statistical measurements such as area, minor-
to-major axis length ratio, filled area to area ratio,
filled area to convex area ratio and number of holes,
and
Sum Freeman chain code.
As shown in Fig. 2(b), the pixels x
1
, x
2
, , x
8
are the 8-
neighbours of p in its 33 template and said to be 8-adjacent to
p. We will use x
i
to denote both the pixel and its value 0 or 1, x
i

is called white or black, accordingly. In the thinned image, the
pixel p is examined for termination point is detected if T
(summation of neighbors of the central pixel p, where p is a
white pixel and its value equal to 1) is equal to 1 as shown in in
the following equation:
1
7
0

i
i
x T

IV. CLASSIFICATION
After the features extraction in the previous section, the
classification process in our recognition system is divided into
two stages. In the first stage, mainly counting the number of
termination points, the digits are classified into six categories.
The relation between the number of the termination points and
possible digits is shown in Table I. In the second stage, other
features such as freeman chain code, orientation, positions,
distances and others, all the digits are recognized.
TABLE I. THE RELATION BETWEEN NUMBER OF TERMINATION
POINTS AND DIGITS

Possible digits
0 1 2 3 4 5 6 7 8 9
N
u
m
b
e
r

o
f

t
e
r
m
i
n
a
t
i
o
n

p
o
i
n
t
s
0
1
2
3
4
5
> 5 Rejected
The implementation of the above mentioned two-staged
classification is depicted in Figure 3. We can obviously notice
the following:
1. The result of the thinned operation appears as blue
colored dots,
2. Each termination point is bounded with a red square
color including its orientation information,
3. The sequence of each digit image as exists in its test
set appears on the top-left corner with blue color, and
4. The classification result appears on the bottom-left
corner with blue color.

Preprocess
Features
Extraction
Classification
Input
image
Classified
image
0
1
2
3
4
5
5
6
7
x
3
x
2
x
1

x
4
p x
0

x
5
x
6
x
7

(a) (b)
(1)
Vol. 2, No.10, 2011

39 | P a g e

Figure 3. The implementation of the two-staged classification process
V. EXPERIEMENTAL RESULTS
Our experiments are performed on MNIST digit database
by using MATLAB 6.5.1 release 13. This database consists of
10,000 gray-scale digit patterns in testing set and 60,000 gray-
scale digit patterns in training set. The size of each digit image
is 2828 pixels. All results are obtained by using 2.40 GHz P4
processor under Windows XP. The processing time of each
single digit is around 0.488 s.
The results of our recognition system tested on the testing
set are summarized in Table II and on the training set are
shown in Table III. It's shown that for each class, there exists a
true positive (TP) means that the digit is correctly classified,
false positive (FP) means that the digit is wrongly classified,
rejected (RJ) means that the digit is not classified or unknown
and reliability (RL) is calculated as:
100 *
(%) 100
(%)
(%)
RJ
TP
RL

TABLE II. THE PERFORMANCE OF THE CLASSIFIERS ON THE
TESTING SET

#
No. of
Images
True Positive
(TP%)
False Positive
(FP%)
Rejected
(RJ%)
Reliability
(RL%)
0 980 95.5102 2.8571 1.6327 97.0955
1 1135 97.0925 2.5551 0.3524 97.4359
2 1032 92.3450 4.6512 3.0039 95.2049
3 1010 94.3564 4.4554 1.1881 95.4909
4 982 94.3992 4.5825 1.0183 95.3704
5 592 90.5405 6.9257 2.5338 92.8943
6 958 91.0230 6.0543 2.9228 93.7635
7 1028 91.1479 6.4202 2.4319 93.4198
8 974 90.5544 8.9322 0.5133 91.0216
9 1009 92.0714 4.9554 2.9732 94.8928
Overall
Performance
%
92.9041 5.2389 1.8570 94.6589
TABLE III. THE PERFORMANCE OF THE CLASSIFIERS ON THE
TRAINING SET
#
True positive False positive Rejection Reliability
No. of
Images
TP%
No. of
Images
FP%
No. of
Images
RJ% RL%
0 5698 96.2175 159 2.6849 65 1.6327 97.2853
1 6498 97.7584 113 1.7000 36 0.3524 98.2907
2 5803 94.2658 246 3.9799 108 3.0039 95.9491
3 5918 94.9615 237 3.8030 77 1.1881 96.1495
4 6433 95.2332 259 3.8342 63 1.0183 96.1297
5 5126 93.0309 253 4.5917 131 2.5338 95.2965
6 7810 94.7011 301 3,6498 136 2.9228 96.2890
7 5150 94.8260 196 3.6089 85 2.4319 96.3337
8 3235 95.2872 119 3.5052 41 0.5133 96.4520
9 5401 94.6713 211 3.6985 93 2.9732 96.2402
57072 2093 835
O
v
e
r
a
l
l

P
e
r
f
o
r
m
a
n
c
e

%

95.0953 3.5056 1.3991 96.4416
As shown in the above tables, the overall performance for
Latin handwritten digit recognition is 92.9041% and 95.0953,
consecutively.
VI. CONCLUSION AND FUTURE WORK
In this paper, we have described a simple approach for
feature extraction and classification for Latin handwritten
recognition. Five feature sets are obtained from the thinned
digit image using the concept of the freeman chain code
template. A two-staged classification is implemented. Firstly,
counting the number of termination points, the digits are
classified into six categories. The relation between the number
of the termination points and possible digits is easily detected.
Secondly, other features such as freeman chain code,
orientation, positions, distances and others are calculated. The
overall performance is 92.9041% for the recognition of 10,000
digit images in the test set and 95.0953% for the recognition of
60,000 digit images in the training set.
In the future work, a tapped operation for disconnected
digits will be tested in the preprocess stage. Furthermore, we
will test the above mentioned approach in the classification of
the formation of the characters in the Arabic language.
REFERENCES
[1] D. Gorgevik, D. Cakmakov and V. Radevski, Handwritten Digit
Recognition by Combining Support Vector Machines Using Rule-Based
Reasoning, Proc. of the 23
rd
International Conference on Information
Technology Interfaces. 19(12) (2001) 139144.
[2] D. Gorgevik and D. Cakmakov, Combining SVM Classifiers for
Handwritten Digit Recognition, Proc. of the 16
th
International
Conference on Pattern Recognition. 3 (2002) 102105.
(2)
Vol. 2, No.10, 2011

40 | P a g e
[3] S. Maji and J. Malik, Fast and Accurate Digit Classification, Technical
Report No. UCB/EECS-2009-159. (2009).
[4] Z. Chi, Z. Lu and F-H. Chan, Multi-Channel Handwritten Digit
Recognition Using Neural Networks, Proc. of IEEE International
Symposium on Circuits and Systems. 1(1997) 625628.
[5] B. Duerr, W. Haettich, H. Tropf and G. Winkler, A combination of
statistical and syntactical pattern recognition applied to classification of
unconstrained handwritten numerals, Pattern Recognition. 12 (1980)
189199.
[6] F. Shah and K. Yousaf, Handwritten Digit Recognition Using Image
Processing and Neural Networks, Proceedings of the World Congress on
Engineering. 1 (2007).
[7] E. Kussul and T. Baidyk, Improved method of handwritten digit
recognition tested on MNIST database, Image and Vision Computing.
22(12) (2004) 971-981.
[8] A. Jain and D. Zongker, Representation and Recognition of Handwritten
Digits Using Deformable Templates, IEEE Trans. Pattern Analysis and
Machine Intelligence. 19(19) (1997) 13861390.
[9] V. Mane and L. Ragha, Handwritten digit recognition using elastic
matching, Proceedings of the International Conference & Workshop on
Emerging Trends in Technology. (2011), pp. 1364.
[10] D. Gorgevik and D. Cakmakov, D, An efficient Three-Stage Classifier
for Handwritten Digit Recognition, Proc. of the 17
th
International
Conference on Pattern Recognition. 4 (2004) 507510.
[11] A. Bellili, M. Gilloux and P. Gallinari, An Hybrid MLP-SVM
handwritten digit recognizer, Proc. of the 6
th
International Conference
on Document Analysis and Recognition. (2001), pp. 2832.
[12] C. Pereira and G. Cavalcanti, Prototype Selection for Handwritten
Connected Digits Classification, The 10
th
International Conference on
Document Analysis and Recognition. (2009), pp. 1021-1025.
[13] R. Ebrahimpour and S. Hamedi, Hand Written Digit Recognition by
Multiple Classifier Fusion based on Decision Templates Approach, In
Proc. of the International Conference on Computer, Electrical, Systems
Science, and Engineering (CESSE). (2009), pp. 245-250.
[14] L. Cheng-Lin, N. Kazuki, H. Sako and H. Fujisawa, Handwritten Digit
Recognition: Benchmarking of State-of-art Techniques, Pattern
Recognition. 36 (2003) 2271-2285.
[15] MNIST Database of Handwritten digits: http://yann.lecun.com/exdb/mni
st/.
[16] L. Lam, S-W. Lee and C. Suen, Thinning methodologies-a
comprehensive survey, IEEE Trans. Pattern Analysis and Machine
Intelligence. 14(9) (1992) 869885.
[17] H. Freeman, On the Encoding of Arbitrary Geometric Configurations,
IRE Trans. Pattern Analysis and Machine Intelligence Electronics and
Computers. 10(1961) 260268.
[18] Kekre, H. B. (2010). Texture Based Segmentation using Statistical
Properties for Mammographic Images. IJACSA - International Journal
of Advanced Computer Science and Applications, 1(5), 102-107.
[19] Malhotra, R. (2011). Software Effort Prediction using Statistical and
Machine Learning Methods. IJACSA - International Journal of
Advanced Computer Science and Applications, 2(1), 145-152.
AUTHOR PROFILE
Ihab Zaqout received the B. S. in Computer Science from
the University of Al-Fateh, Libya, in 1987 and the M.S.
degree in Computer Science from Jordan University,
Jordan in 2000 and the Ph.D. in Computer Science from
the University of Malaya, Malaysia in 2006. He is
currently at the Dept. of Information Technology, Al-
Azhar University - Gaza, Palestine as an assistant professor. His main research
interests include image processing, pattern recognition, data mining and
machine learning.

Vol. 2, No. 10, 2011

41 | P a g e
Plant Leaf Recognition using Shape based Features
and Neural Network classifiers

Jyotismita Chaki
School of Education Technology
Jadavpur University
Kolkata, India
Ranjan Parekh
School of Education Technology
Jadavpur University
Kolkata, India

AbstractThis paper proposes an automated system for
recognizing plant species based on leaf images. Plant leaf images
corresponding to three plant types, are analyzed using two
different shape modeling techniques, the first based on the
Moments-Invariant (M-I) model and the second on the Centroid-
Radii (C-R) model. For the M-I model the first four normalized
central moments have been considered and studied in various
combinations viz. individually, in joint 2-D and 3-D feature
spaces for producing optimum results. For the C-R model an
edge detector has been used to identify the boundary of the leaf
shape and 36 radii at 10 degree angular separation have been
used to build the feature vector. To further improve the accuracy,
a hybrid set of features involving both the M-I and C-R models
has been generated and explored to find whether the combination
feature vector can lead to better performance. Neural networks
are used as classifiers for discrimination. The data set consists of
180 images divided into three classes with 60 images each.
Accuracies ranging from 90%-100% are obtained which are
comparable to the best figures reported in extant literature.
Keywords-plant recognition; moment invariants; centroid-radii
model; neural network; computer vision.
I. INTRODUCTION
Plants play an important role in our environment. Without
plants there will be no existence of the earths ecology. But in
recent days, many types of plants are at the risk of extinction.
To protect plants and to catalogue various types of flora
diversities, a plant database is an important step towards
conservation of earths biosphere. There are a huge number of
plant species worldwide. To handle such volumes of
information, development of a quick and efficient classification
method has become an area of active research. In addition to
the conservation aspect, recognition of plants is also necessary
to utilize their medicinal properties and using them as sources
of alternative energy sources like bio-fuel. There are several
ways to recognize a plant, like flower, root, leaf, fruit etc. In
recent times computer vision methodologies and pattern
recognition techniques have been applied towards automated
procedures of plant recognition.
The present paper proposes a scheme for automated
recognition of three types of plant species by analyzing shape
features from digital images of their leaves. The organization of
the paper is as follows: section 2 provides an overview of
related work, section 3 outlines the proposed approach with
discussions on overview, feature computation and classification
schemes, section 4 provides details of the dataset and
experimental results obtained and section 5 provides the overall
conclusion and the scope for future research.
II. PREVIOUS WORK
Many methodologies have been proposed to analyze plant
leaves in an automated fashion. A large percentage of such
works utilize shape recognition techniques to model and
represent the contour shapes of leaves, however additionally,
color and texture of leaves have also been taken into
consideration to improve recognition accuracies. One of the
earliest works [1] employs geometrical parameters like area,
perimeter, maximum length, maximum width, elongation to
differentiate between four types of rice grains, with accuracies
around 95%. Use of statistical discriminant analysis along with
color based clustering and neural networks have been used in
[2] for classification of a flowered plant and a cactus plant. In
[3] the authors use the Curvature Scale Space (CSS) technique
and k-NN classifiers to classify chrysanthemum leaves. Both
color and geometrical features have been reported in [4] to
detect weeds in crop fields employing k-NN classifiers. In [5]
the authors propose a hierarchical technique of representing
leaf shapes by first their polygonal approximations and then
introducing more and more local details in subsequent steps.
Fuzzy logic decision making has been utilized in [6] to detect
weeds in an agricultural field. In [7] the authors propose a two-
step approach of using a shape characterization function called
centroid-contour distance curve and the object eccentricity for
leaf image retrieval. The centroid-contour distance (CCD)
curve and eccentricity along with an angle code histogram
(ACH) have been used in [8] for plant recognition. The
effectiveness of using fractal dimensions in describing leaf
shapes has been explored in [9].
In contrast to contour-based methods, region-based shape
recognition techniques have been used in [10] for leaf image
classification. Elliptic Fourier harmonic functions have been
used to recognize leaf shapes in [11] along with principal
component analysis for selecting the best Fourier coefficients.
In [12] the authors propose a leaf image retrieval scheme based
on leaf venation for leaf categorization. Leaf venations are
represented using points selected by the curvature scale scope
corner detection method on the venation image and categorized
by calculating the density of feature points using non
parametric estimation density. In [13] 12 leaf features are
extracted and orthogonalized into 5 principal variables which
Vol. 2, No. 10, 2011

42 | P a g e
consist of the input vector of a neural network (NN). The NN is
trained by 1800 leaves to classify 32 kinds of plants with
accuracy greater than 90%. NNs have also been used in [14] to
classify plant based on parameters like size, radius, perimeter,
solidity and eccentricity. An accuracy of about 80% is reported.
Wavelet and fractal based features have been used in [15] to
model the uneven shapes of leaves. Texture features along with
shape identifiers have been used in [16] to improve recognition
accuracies. Other techniques like Zernike moments and Polar
Fourier Transform have also been proposed [17] for modeling
leaf structures. An accuracy of 64% has been reported. In [18]
authors propose Hybrid Image Segmentation Algorithm for
Leaf Recognition and Characterization.
A new approach that combines a thresholding method and
H-maxima transformation based method is proposed to extract
the leaf veins. Compared with other methods, experimental
results show that this combined approach is capable of
extracting more accurate venation modality of the leaf for the
subsequent vein pattern classification. In [19] authors propose
Guiding Active Contours for Tree Leaf Segmentation and
Identification. Combining global shape descriptors given by the
polygonal model with local curvature-based features, the leaves
are classified over nearly 50 tree species. Finally in [20] a
combination of all image features viz. color, texture and shape,
have been used for leaf image retrieval, with a reported
accuracy of 97.9%.
III. PROPOSED APPROACH
The present paper proposes a scheme for automated
detection of 3 classes of plant category by analyzing shapes
obtained from a collection of their leaf images, using features
based on Moment-Invariants and Centroid-Radii approaches,
with various types of neural network classifiers:
A. Moment Invariants (M-I)
M-K Hu [21] proposes 7 moment features that can be used
to describe shapes and these are invariant to rotation,
translation and scaling. For a digital image, the moment of a
pixel
( , ) P x y
at location
( , ) x y
is defined as the product of the
pixel value with its coordinate distances i.e.
. . ( , ) m x y P x y =
.
The moment of the entire image is the summation of the
moments of all its pixels. More generally the moment of order
( , ) p q
of an image
( , ) I x y
is given by

[ ( , )]
p q
pq
x y
m x y I x y =
(1)
Based on the values of p and q the following are defined :
0 0
00
1 0
10
0 1
01
1 1
11
2 0 2
20
0 2
02
[ . ( , )] [ ( , )]
[ . ( , )] [ . ( , )]
[ . ( , )] [ . ( , )]
[ . ( , )] [ . ( , )]
[ . ( , )] [ . ( , )]
[ . ( , )]
x y x y
x y x y
x y x y
x y x y
x y x y
x y
m x y I x y I x y
m x y I x y x I x y
m x y I x y y I x y
m x y I x y xy I x y
m x y I x y x I x y
m x y I x y
= =
= =
= =
= =
= =
=

2
2 1 2
21
1 2 2
12
3 0 3
30
0 3 3
03
[ . ( , )]
[ . ( , )] [ . ( , )]
[ . ( , )] [ . ( , )]
[ . ( , )] [ . ( , )]
[ . ( , )] [ . ( , )]
x y
x y x y
x y x y
x y x y
x y x y
y I x y
m x y I x y x y I x y
m x y I x y xy I x y
m x y I x y x I x y
m x y I x y y I x y
=
= =
= =
= =
= =

(2)

The first four Hu invariant moments which are invariant to
rotation are defined as follows

1 20 02
2 2
2 20 02 11
2 2
3 30 12 21 03
2 2
4 30 12 21 03
( ) (2 )
( 3 ) (3 )
( ) ( )
m m
m m m
m m m m
m m m m
= +
= +
= +
= + + +
(3)
To make the moments invariant to translation the image is
shifted such that its centroid coincides with the origin of the
coordinate system. The centroid of the image in terms of the
moments is given by:

10
00
01
00
c
c
m
x
m
m
y
m
=
=
(4)
Then the central moments are defined as follows:

[( ) ( ) ( , )]
p q
pq c c
x y
x x y y I x y =
(5)
Vol. 2, No. 10, 2011

43 | P a g e

To compute Hu moments using central moments the
terms in equation (2) need to be replaced by terms. It can be
verified that
00
= m
00
,
10
= 0 =
01
.
To make the moments invariant to scaling the moments are
normalized by dividing by a power of
00
. The normalized
central moments are defined as below

00
, where 1
( ) 2
pq
pq
p q
M
e
+
= = +
(6)
B. Centroid-Radii model (C-R)
In [22] K. L. Tan et al. proposes the centroid-radii model
for estimating shapes of objects in images. A shape is defined
to be an area of black on a background of white. Each pixel is
represented by its color (black or white) and its x-y co-
ordinates on the canvas. The centroid is located at the position
(Cx, Cy) which are respectively, the average of the x and y co-
ordinates for all black pixels. The boundary of a shape consists
of a series of boundary points. A boundary point is a black
pixel with a white pixel as its neighbor. A radius is a straight
line joining the centroid to a boundary point. In the centroid-
radii model, lengths of a shapes radii from its centroid at
regular intervals are captured as the shapes descriptor using
the Euclidean distance. More formally, let be the regular
interval (measured in degrees) between radii (Figure 1). Then,
the number of intervals is given by k =360/. All radii lengths
are normalized by dividing with the longest radius length from
the set of radii lengths extracted. Furthermore, without loss of
generality, suppose that the intervals are taken clockwise
starting from the x-axis direction. Then, the shape descriptor
can be represented as a vector :

0 2 ( 1)
{ , , ,..., }
k
S r r r r
u u u
= (7)
Here r
i
, 0 i (k-1) is the (i+1)-th radius from the
centroid to the boundary of the shape. With sufficient number
of radii, dissimilar shapes can be differentiated from each
other.
Figure 1. Centroid-radii approach
IV. EXPERIMENTATIONS AND RESULTS
Experimentations are performed by using 180 leaf images
from the Plantscan database available at the URL:
http://imedia-ftp.inria.fr:50012/Pl@ntNet/plantscan_v2/ The
dataset is divided into 3 classes A (Pittosporum Tobira), B
(Betula Pendula), C (Cercis Siliquastrum) each consisting of 60
images. Each image is 350 by 350 pixels in dimensions and in
JPG format. A total of 90 images are used as the Training set
(T) and the remaining 90 images as the Testing set (S). Sample
images of each class are shown below.
Figure 2. Samples of leaf images belonging to 3 classes
For computing recognition rates, comparisons between
training and testing sets are done using neural network (Multi-
layer perceptrons : MLP) with feed-forward back-propagation
architectures. The legends used in this work are listed in Table
I.
TABLE I. LEGENDS
Legend Meaning Legend Meaning
AT Class-A Training set AS Class-A Testing set
BT Class-B Training set BS Class-B Testing set
CT Class-C Training set CS Class-C Testing set
A. Moment-Invariants (M-I) Representations
The first 4 normalized central moments M1, M2, M3, M4
of each image of the training and testing sets were computed as
per equation (6). Various combination of the features in
individual and joint configurations were fed to neural network
(NN) classifiers to study which combinations produce the best
results
1) Individual Features:
Individual features values M1, M2, M3, M4 for the training
and testing images for the 3 classes are first used. Results are
summarized in Table II below. The first column depicts the
feature used, the second column shows the neural network
configuration (NNC) viz. 1-3-3 indicates 1 input unit (for the
individual feature), 3 units in the hidden layer and 3 units in the
output layer (corresponding to the 3 classes to be
distinguished).
Vol. 2, No. 10, 2011

44 | P a g e
The third, fourth and fifth columns indicate the percentage
recognition accuracies for the three classes, the sixth column
provides the overall accuracy for the three classes and the last
column indicates the best Mean Square Error (MSE) obtained
during the training phase of the NNs.
TABLE II. ACCURACY USING INDIVIDUAL M-I FEATURES
F NNC A B C O MSE
M1 1-3-3
70
(21/30)
96.7
(29/30)
100
(30/30)
88.9
(80/90)
0.06
M2 1-3-3 73.3 86.7 20 60 0.17
M3 1-3-3 10 100 70 60 0.15
M4 1-3-3 86.7 46.7 50 61.1 0.16
Table II indicates that of the individual moment values M
1

provides the best results of 88.9 %. The number of images (out
of 30) correctly identified, are also indicated in parenthesis.
The corresponding NN output for the 3 classes, is shown
below in Figure 3. The class files are arranged sequentially i.e.
the first 30 files belong to Class-A the next 30 to Class-B and
the last 30 to Class-C.
2) Joint Features:
To improve upon the results obtained using individual
features, joint features are next considered in 2-D feature
spaces i.e. M
1
-M
2
, M
1
-M
3
, M
1
-M
4
. M
1
is kept common since it
is seen to produce the best accuracies. Results are summarized
in Table III.
TABLE III. ACCURACY USING JOINT 2-D M-I FEATURES
F NNC A B C O MSE
M1-M2 2-3-3 90 100 93.3 94.4 0.058
M1-M3 2-3-3
90
(27/30)
96.6
(29/30)
100
(27/30)
95.5
(86/90)
0.053
M1-M4 2-3-3 86.6 96.6 93.3 92.2 0.068

Table III indicates that of the joint 2-D feature values M
1
-
M
3
provides the best results of 95.5 %. The corresponding NN
output for the 3 classes, is shown below in Figure 4.
To conclude analysis of joint spaces, features are also
considered in 3-D feature spaces : M
1
-M
2
-M
3
, M
1
-M
2
-M
4
, M
1
-
M
3
-M
4
. Results are summarized on Table IV below.
TABLE IV. ACCURACY USING JOINT 3-D M-I FEATURES
F NNC A B C O MSE
M1-M2-M3 3-30-3 83.3 100 93.3 92.2 0.028
M1-M2-M4 3-30-3 90 86.7 90 88.9 0.005
M1-M3-M4 3-30-3
90
(27/30)
100
(30/30)
90
(27/30)
93.3
(84/90)
0.005

Table IV indicates that of the joint 3-D feature values M
1
-
M
3
-M
4
provides the best results of 93.3 %. The corresponding
NN output for the 3 classes, is shown in Figure 5.
Figures 6 and 7 shows the variation of the best features M
1
,
(M
1
-M
3
), (M
1
-M
3
-M
4
) over the training and testing files for the
3 classes.

Figure 3. NN output for M1

Figure 4. NN output for M1-M3

Figure 5. NN output for M1-M3-M4

Vol. 2, No. 10, 2011

45 | P a g e

Figure 6. Variation of individual M1 and M1-M3 features in 2-D space

Figure 7. Variation of M1-M3-M4 features in 3-D space
B. Centroid-Radii (C-R) Representations
Each image is converted to binary form and the Canny edge
detector is used to identify its contour. Its centroid is computed
from the average of its edge pixels.
Corresponding to each edge pixel the angle it subtends at
the centroid is calculated and stored in an array along with its
x- and y- coordinate values. From the array 36 coordinate
values of edge pixels which join the centroid at 10 degree
intervals from 0 to 359 degrees are identified. The radii length
of joining these 36 points with the centroid are calculated using
the Euclidean distance and the radii lengths are normalized to
the range [0,1]. For each leaf image 36 such normalized lengths
are stored in an ordered sequence. Figure 8 shows a visual
representation of a leaf image, the edge detected version, the
location of the centroid and edge pixels, and the normalized
radii vector. The average of the 36 radii lengths for each image
of each class both for the training and testing sets, is plotted in
Figure 9, which depicts the overall feature range and variation
for each class.
Classes are discriminated using NN. The results are
summarized in Table V. An overall accuracy of 100% is
achieved. The NN convergence plot and output are shown in
Figure 10. Convergence takes place in 38280 epochs with an
MSE of 0.005.
TABLE V. ACCURACY USING C-R FEATURES
F NNC A B C O MSE
36-element
C-R vector
36-
30-3
100
(30/30)
100
(30/30)
100
(30/30)
100
(90/90)
0.005

Figure 8. Interface for C-R computations

Figure 9. Variation of mean radii length for 3 classes
C. Hybrid Representation
To study recognition accuracy of a hybrid set of features,
two combinations of C-R and M-I features were used : (1) 38-
element vector : 36-elements from C-R model and 2 elements
(M1, M3) from M-I model (2) 39-element vector : 36-elements
from C-R model and 3 elements (M1, M3, M4) from M-I
model. The combination vectors are fed to an appropriate NN
and results in each case are summarized below in Table VI.
TABLE VI. ACCURACY USING HYBRID FEATURES

F NNC A B C O MSE
38-element
hybrid
vector
38-30-3
100
(30/30)
100
(30/30)
100
(30/30)
100
(90/90)
0.005
39-element
hybrid
vector
39-30-3 100 100 93.3 97.7 0.005
Vol. 2, No. 10, 2011

46 | P a g e
The overall accuracy is observed to be 100% using the 38-
element hybrid vector, but convergence takes place much
quicker requiring only 20706 epochs for an MSE of 0.005, in
contrast to 38280 epochs when using only the C-R vector. The
convergence plot and NN output is shown in Figure 11.
Regarding the system implementation, computation of 36-
radii vector for 30 images takes around 35 seconds, while
computation of the first four moments for 30 images takes
around 300 seconds, on a 2.66 GHz P4 system with 1 GB
RAM. NN convergence for the 38-element hybrid vector takes
around 2 minutes for a training set of 30 images. The system
has been observed to be stable for all the 3 categories of the
leaf images displaying almost the same timings over the entire
dataset.

Figure 10. NN convergence and output for C-R vector
V. ANALYSIS
Automated discrimination between three leaf shapes was
done using a variety of approaches to find the optimum results.
The study reveals that for M-I approach joint features in
general provide better accuracies than individual features.
Accuracies based on C-R method are better than individual M-I
features and comparable to joint M-I features.

Figure 11. NN convergence and output for 38-e hybrid vector
Performance was found to improve when C-R features were
combined with M-I features in a hybrid feature space, the
accuracy of 100% being same as that for only the C-R vector,
but requiring almost half the number of epochs. Accuracy
results obtained using different methods are summarized in
Table VII.
TABLE VII. ACCURACY USING VARIOUS FEATURES
F M-I I M-I J2 M-I J3 C-R H
% Accuracy 88.9 95.5 93.3 100 100
To put the above results in perspective with the state of the
art, the best results reported in [8] is a recall rate of 60% for
discrimination of chrysanthemum leaves from a database of
1400 color images. Accuracy for classification for 10 leaf
categories over 600 images is reported to be 82.33% in [10].
Overall classification accuracy reported in [11] for 4 categories
of leaf images obtained during three weeks of germination, is
around 90%. Accuracy reported in [13] for classification of 32
leaf types from a collection of 1800 images is around 90%. An
overall classification of 80% is reported in [14] for identifying
two types of leaf shapes from images taken using different
frequency bands of the spectrum.

Vol. 2, No. 10, 2011

47 | P a g e
Best accuracies reported in [17] are around 93% using Polar
Fourier Transforms. Results reported in [19] are in the region
of 80% for classifying 50 species. Accuracies of around 97%
have been reported in [20] for a database of 500 images. It
therefore can be said that the accuracies reported in the current
paper are comparable to the best results reported in extant
literature. It may however be noted that in many of the above
cases color and geometrical parameters have also been
combined with shape based features to improve results, while
the current work is based solely on shape characteristics.
VI. CONCLUSIONS AND FUTURE SCOPES
This paper proposes an automated system for plant
identification using shape features of their leaves. Two shape
modeling approaches are discussed: one technique based on
invariant-moments model and the other on centroid-radii
model, and the two are compared with regard to classification
accuracies. Such automated classification systems can prove
extremely useful for quick and efficient classification of plant
species. The accuracy of the current proposed approach is
comparable to those reported in contemporary works. A salient
feature of the current approach is the low-complexity data
modeling scheme used whereby dimensionality of the feature
vectors are typically below 40.
Future work would involve research along two directions:
(1) combining other shape based techniques like Hough
transform and Fourier descriptors, and (2) combining color and
texture features along with shape features for improving
recognition accuracies.
REFERENCES
[1] N. Sakai, S. Yonekawa, and A. Matsuzaki, Two-dimensional image
analysis of the shape of rice and its applications to separating varieties,
Journal of Food Engineering, vol 27, 1996, pp. 397-407.
[2] A. J. M. Timmermans, and A. A. Hulzebosch, Computer vison system
for on-line sorting of pot plants using an artificial neural network
classifier, Computers and Electronics in Agriculture, vol. 15, 1996, pp.
41-55.
[3] S. Abbasi, F. Mokhtarian, and J. Kittler, Reliable classification of
chrysanthemum leaves through curvature scale space, Lecture Notes in
Computer Science, vol. 1252, 1997, pp. 284-295.
[4] A. J. Perez, F. Lopez, J. V. Benlloch, and S. Christensen, Color and
shape analysis techniques for weed detection in cereal fields,
Computers and Electronics in Agriculture, vol. 25, 2000, pp. 197-212.
[5] C. Im, H. Nishida, and T. L. Kunii, A hierarchical method of
recognizing plant species by leaf shapes, IAPR Workshop on Machine
Vision Applications, 1998, pp. 158-161.
[6] C-C Yang, S. O. Prasher, J-A Landry, J. Perret, and H. S. Ramaswamy,
Recognition of weeds with image processing and their use with fuzzy
logic for precision farming, Canadian Agricultural Emgineering, vol.
42, no. 4, 2000, pp. 195-200.
[7] Z. Wang, Z. Chi, D. Feng, and Q. Wang, Leaf image retrieval with
shape feature, International Conference on Advances in Visual
Information Systems (ACVIS), 2000, pp. 477-487.
[8] Z. Wang, Z. Chi, and D. Feng, Shape based leaf image retrieval, IEEE
Proceedings on Vision, Image and Signal Processing (VISP), vol. 150,
no.1, 2003, pp. 34-43.
[9] J. J. Camarero, S. Siso, and E.G-Pelegrin, Fractal dimension does not
adequately describe the complexity of leaf margin in seedlings of
Quercus species, Anales del Jardn Botnico de Madrid, vol. 60, no. 1,
2003, pp. 63-71.
[10] C-L Lee, and S-Y Chen, Classification of leaf images, 16
th
IPPR
Conference on Computer Vision, Graphics and Image Processing
(CVGIP), 2003, pp. 355-362.
[11] J. C. Neto, G. E. Meyer, D. D. Jones, and A. K. Samal, Plant species
identification using elliptic Fourier leaf shape analysis, Computers and
Electronics in Agriculture, vol. 50, 2006, pp. 121-134.
[12] J-K Park, E-J Hwang, and Y. Nam, A vention based leaf image
classification scheme, Alliance of Information and Referral Systems,
2006, pp. 416-428.
[13] S. G. Wu, F. S. Bao, E. Y. Xu, Y-X Wang, Y-F Chang, and Q-L Xiang,
A leaf recognition algorithm for plant classification using probabilistic
neural network, The Computing Research Repository (CoRR), vol.1,
2007, pp. 11-16.
[14] J. Pan, and Y. He,Recognition of plants by leaves digital image and
neural network, International Conference on Computer Science and
Software Engineering, vol 4, 2008, pp. 906 910.
[15] Q-P Wang, J-X Du, and C-M Zhai, Recognition of leaf image based on
ring projection wavelet fractal feature, International Journal of
Innovative Computing, Information and Control, 2010, pp. 240-246.
[16] T. Beghin, J. S. Cope, P. Remagnino, and S. Barman, Shape and texture
based plant leaf classification, International Conference on Advanced
Concepts for Intelligent Vision Systems (ACVIS), 2010, pp. 345-353.
[17] A. Kadir, L.E. Nugroho, A. Susanto, and P.I. Santosa, A comparative
experiment of several shape methods in recognizing plants,
International Journal of Computer Science & Information Technology
(IJCSIT), vol. 3, no. 3, 2011, pp. 256-263
[18] N. Valliammal, and S. N. Geethalakshmi, Hybrid image segmentation
algorithm for leaf recognition and characterization, International
Conference on Process Automation, Control and Computing (PACC),
2011, pp. 1-6.
[19] G. Cerutti, L. Tougne, J. Mille, A. Vacavant, and D. Coquin, Guiding
active contours for tree leaf segmentation and identification, Cross-
Language Evaluation Forum (CLEF), Amsterdam, Netherlands, 2011.
[20] B. S. Bama, S. M. Valli, S. Raju, and V. A. Kumar, Conten based leaf
image retrieval using shape, color and texture features, Indian Journal
of Computer Science and Engineering, vol. 2, no. 2, 2011, pp. 202-211.
[21] M-K Hu, Visual pattern recognition by moment invariants, IRE
Transactions on Information Theory, 1962, pp. 179-187.
[22] K-L Tan, B. C. Ooi, and L. F. Thiang, Retrieving similar shapes
effectively and efficiently, Multimedia Tools and Applications, vol. 19,
2003, pp. 111-134
AUTHORS PROFILE
Smt. Jyotismita Chaki is a Masters (M.Tech.) research scholar at the School of
Education Technology, Jadavpur University, Kolkata, India. Her
research interests include image processing and pattern recognition.
Dr. Ranjan Parekh is a faculty at the School of Education Technology,
Jadavpur University, Kolkata, India. He is involved with teaching
subjects related to multimedia technologies at the post-graduate level.
His research interests include multimedia databases, pattern recognition,
medical imaging and computer vision. He is the author of the book
Principles of Multimedia published by McGraw-Hill, 2006.

Vol. 2, No. 10, 2011

48 | P a g e
Integrated Information System for reserving rooms in
Hotels

Dr. Safarini Osama
IT Department
University of Tabuk,
Tabuk, KSA
[email protected]

Abstract It is very important to build new and modern flexible
dynamic effective compatible reusable information systems
including database to help manipulate different processes and
deal with many parts around it. One of these is managing the
room reservations for groups and individual, to focus necessary
needs in hotels and to be integrated with accounting system. This
system provides many tools can be used in taking decision.
Keywords: IS - Information System; MIS -Management
Information System; DFD - Data Flow Diagram; ER- Entity
Relationship; DBMS - Database Management system.
I. INTRODUCTION
Since tourism is now taking advantage in all over the
world, so the role of hotels is coming greater than before, from
this point of view our system gains its importance. We had
followed the standards of building an information system [1],
which is analysis, design, and implementation.
Hotels are big organizations that take place in the market,
so using manual system may lead to more and more errors,
because people forgot, and their thinking is limited, so we
think that using a computerized system is more efficient,
especially that we are living in information age.
Reservation is the most important part of any hotel system,
this process should be efficient. So it can serve the commercial
goals of the hotel. The most important entity in the
Reservation process is the Rooms, reserving Rooms in a hotel
should be dynamic, and the whole process should be done in
an effective way to maximize the hotel profits.
We have two types of customers in the hotel; individual
customers and group customers, and the person who is
responsible for reservation is the front office manager. Group
customers have higher priority over individual customers in
high season, because group customers are more stable. So the
front office manager should take decisions about the
reservation, these decisions can be supported by the
computerized system. This system provides many tools that
the front office manager could use in taking his decision for
allocating rooms and reserve for groups and taking right
decisions.
Note that this system can't operate alone in the hotel,
because it concentrates mainly in reservation process, the
system needs an accounting system and a management
information system, and so it can serve all the needs of the
hotel.
We have tried to make it possible to connect our system to
other systems (Accounting, MIS) though two screens which
are; Accounting screen which deals with entering bills for a
customer, and recipient screen which deals with adding a
recipient for the hotel. A very important part that we include in
our system is the report generator, which enables the manager
to have any information about the customers in the hotel. We
tried to improve the reusability of our system through setting
screen which lets the user change the main settings such as
name of the hotel, colors of the background of the screens, and
hotel logo.
A. Similar Projects
1) Fidelio Hotel System
One of the most famous hotels systems is Fidelio, which
was developed by Microsoft; it can be considered as a
complete hotel system, since it has a hotel accounting system
and a management information system in addition to
reservation system. It was developed in 1988, but in those
days computers was in its first stages to the commercial
business and big organizations, so Fidelio wasn't famous at his
first appearance. Over and over when computers take place in
the market, big organizations such as hotels start using
computerized systems, and then Fidelio starts to be famous.
Fidelio reservation system divides the reservation process
into two types, Group and Individual reservation. Both the
recipient and the reservation manager can reserve for any type.
The reservation information and resident information are
entered from the same screen.
When allocating rooms for customers in hotel, you can
click on allocate button to show a small screen containing the
empty rooms. In other words you can fill all information for
customers in the same screen whether this customer was an
individual customer or a group customer. As for accounting,
you enter other screens to add the customer bills, this screen
contains a list of valued facilities, and you can change the cost
from this screen. There is no ability in Fidelio to change the
color of the screen or the fonts, also the name of hotel in not
shown in the main screen.
Fidelio Reservation System contains some static reports
about the customers in hotel. These reports contain static data
Vol. 2, No. 10, 2011

49 | P a g e
about customers such as full name and room numbers. You
cannot change it by removing or adding new fields.
The user can change his password and nickname from the
system that the reservation manager gave it to him. The
system doesn't provide the manager with any kind of statistics
about the situation of the rooms in the hotel. A reservation
manager screen contains two main components. The first is
adding recipient information and assign him a shift and sub-
manager property. The second component is adding new room
to hotel and converts an existing room to maintenance state.
II. DISCUSSION
The designed Information system begins with a login
screen Fig.1

Fig.1 Login Screen
This screen is the first screen in the system which allows
the authorized user to access the system. Our project is talking
mainly about the reservation process, and components needed
to connect the reservation system with other systems needed
by the hotel.
First, we divide the reservation process into two types,
which are groups reservation and individuals reservation,
this division was done so the user will not be confused, and
the information needed from each entity differ from the other.
The reservation process is provided with search
capabilities, allocating rooms for the customer is one of the
most important components of the system which enables the
front office manager and recipient to allocate empty rooms for
the customers. The Main menu is illustrated in fig.2

Fig.2 Main Screen
A. The setting screen:
This screen allows the user to modify some settings such
as the color and the hotel name and the hotels logo (see fig3.)

Fig. 3 Settings Screen
B. Chart Screen (fig. 4)
This screen allow the user to enter only the date from a
certain date to a certain date, then by pressing the View
button it will give the user a pie chart showing him the every
type rooms in the hotel and its partition comparing with other
room types.
The user will get also some additional numeric data.

Fig.4 Chart Screen
C. System Design and Implementation
We have tried to follow the standard phases of building a
system, and our main aim is to design a user friendly system
interface that helps users to finish their work in an efficient
way and the lowest effort, also we tried to add many features
and functions to ease the work process. Hotel systems contain
a dynamic decision for the work to be finished, so we tried to
divide the project into entities according to DFD (data flow
diagram). We choose tab page design so the user will feel
comfort with working with the system.
D. The Accounting System
The system does not include the accounting features in a
detailed way, because hotels have its own systems that deal
with accounting perspective. But in our analysis for the system
we found that some models should be involved in the
reservation system in order to connect this system with the
accounting system of the hotel.
The most important model for doing what mentioned
Vol. 2, No. 10, 2011

50 | P a g e

Log
Shift
Recepiant
Res_Type
Res_group_Type
Customer_Group
Supliers
Res_Individuals
Res_Group
Res_Group_Room
Customer_Individual
Room_state
Rooms
Room_Type
Room_status
Room_Account
Temp
above is the customer accounts, and enabling the recipient to
add bills (fig.5) for the customers depending on their rooms
numbers.
We made a single screen for adding bills to the customers,
the recipient take the bill and choose one facility form the
hotel facilities, and enter the amount of the bill, all what he
have to do is to give the customer room number, also he can
print a bill for the customer at any time while the customer is
in the hotel.

Fig. 5 Accounting Screen
E. Reports
The reports are considered as a very important component
of any system, so we define two categories of reports, which
are; static reports and generated reports.
Since the manager takes place in taking decision in the
hotel, and the reports is one of the most important tools that
enable him to do that, so we concentrate on this component.
There is a static report on each screen; we tried to put the most
common report on each tab page for every screen. Also we
design a report generator that enables the user to have dynamic
reports types and control the data to be included in the report.
Our concentration on the reports is for the following
reasons;
The reports is an essential part of any system,
Helps the managers to take decision about the hotel
Ease the process of retrieving information about the
customers in the group and also the groups.
F. Implementation Decision
We intended to use Oracle8 [2] and Developer6 which
include Forms6i and Report6i; we thought that it's powerful
for implementing the proposed system.
Oracle contains many features that make it over choice for
implementing the project. The main features are:
1) Oracle is a database Management System, which helps
in implementation of enterprise systems.
2) Oracle supports a stable database with minimum
percentage of crash.
3) It helps in designing powerful database tables and
schemes.
Also, the packages included in Oracle [3], such as Form
Builder and Report Builder provides good GUI tools in order
to ease the process of prototyping and designing interfaces in a
user-friendly manner [4]. We had used ER diagram for
database design fig. 6, because it is powerful tool for database
design [5] and scheme elicitation.
Fig. 6 ER diagram
G. System Features
1) Fast and Easy to use
From a system point of view nothing is more important to
the quality of direct customer service (reservations, check-
in/out etc.) than speed and accuracy of response. Our hotel
system challenges overcome all other hotel systems in speed
and in ease of use.
2) Flexible Reservation and Enquiry Procedure
As much or as little information that any Enquirer offers
can be entered in whatever orders it is given. As soon as the
name is entered, be it a previous guest or Enquirer, a corporate
contact or travel agent, a search will rapidly present their
details including e.g. negotiated rates and preferred room
numbers. The speed of this name search is of critical
importance if it is to be effective - firstly as an aid to prompt
customer service and recognition and secondly to ensure the
growth of a 'clean' accurate database without record
duplication. Our hotel fulfills these criteria.
3) Automatic Report Generator
For any needed Enquirer, simply enter their date and make
some choices to have quickly and accurately completed in
front of your eyes a report automatically created. The benefits
are faster customer service, accurate information and reduced
mailing costs. (No special hardware, only a standard PC is
required).
4) On-Screen Room Chart
The Room Chart provides availability and occupancy
information in a familiar format for those transferring from a
manual front office system. Our system is designed to be as
friendly and familiar to the first-time computer user as
possible, which also makes re-training for experienced users
relatively quick and easy.

Vol. 2, No. 10, 2011

51 | P a g e
5) Accounts
The majority of data transfer to accounts is usually
between the front office (facility) and the accounting, which
are totally integrated with our hotel. Finally our hotel System
hopes to provide Travelogue Hotels and Resorts with faster
and better guest service, including up-to-date history
information. The system can provide increased revenues
through tighter database control, as well as forecasting data
and reports to help make smarter management decisions.
6) The Front Office Features
Front office gives the user the ability to enter recipient
employees and give them their nick-names and passwords,
which can be changed by the recipient later on. Also the
search ability for recipients is available.
For new rooms its cost, type, and number can be done
through the front office. To add new facility and its cost, can
be done through the front office. There is the ability to modify
the discount rate and the tax percentage.
Also in the front office there is the ability to the user to
view a semi analysis on the usability of the rooms according to
a given period of time, it views the data at two statistical ways
the first one graphical way by showing a pie chart for rooms
for specified type, the second one by showing a table with
accurate number for rooms.
III. CONCLUSION
A. Comparison with Fidelio System
As we mentioned before, Fidelio is a complete hotel
system, which deals with accounting and management
information systems, in addition of reservation system. So our
comparison, as shown in the following table 1, will be
between over system and Fidelio reservation system.
As a main conclusion is the efficiency in generating
reports, friendly user interfaces, making decision, enough
stable system as we intended to use Oracle8 and Developer6.
Previously discussed system features can be considered as
conclusions.
B. Future Scope
The future plans are to get our developed information
system more integrated with Accounting and Management
systems.
REFERENCES
[1] Principles of Information Systems, 10th Edition, Ralph M. Stair, George
Reynolds; ISBN-10: 0538478292 ISBN-13: 9780538478298, 704
Pages CB 2012 Published
[2] Mastering Oracle PL/SQL: Practical Solutions Copyright 2004 by
Connor McDonald, with Chaim Katz, Christopher Beck, Joel R.
Kallman, and David C. Knox, ISBN (pbk): 1-59059-217-4
[3] Oracle_Database_10g_-_The_Complete_Reference_
_McGraw_Hill_Osborne.pdf 7935 KB
http://rapidshare.com/files/11822933...ll_Osborne.pdf
[4] Programming with Microsoft Visual Basic 6.0: an object-oriented
approach : comprehensive , Edition illustrated; Michael V. Ekedahl,
William Arthur Newman; Course Technology, 1999 - 720 pages; ISBN
0760010765, 9780760010761.
[5] Database Design for Mere Mortals: A Hands-On Guide to Relational
Database Design (2nd Edition) [Paperback] 611 pages, Michael J.
Hernandez Publisher: Addison-Wesley Professional; 2 edition (March
13, 2003) ISBN-10: 0201752840 ISBN-13: 978-0201752847.
[6] Dhindsa, K. S. (2011). Modelling & Designing Land Record
Information System Using Unified Modelling Language. IJACSA -
2(2), 26-30.

TABLE 1 COMPARISON BETWEEN OUR SYSTEM WITH FIDELIO SYSTEM

AUTHORS PROFILE
Dr. Osama Ahmad Salim Safarini had finished his
PhD. from The Russian State University of Oil
and Gaz Named after J. M. Gudkin,
Moscow, 2000, at a Computerized-Control
Systems Department.
He obtained his BSC and MSC in Engineering and
Computing Science from Odessa Polytechnic
National State University in Ukraine 1996. He
worked in different universities and

SYSTEM
ITEM
Our System Fidelio System
Group
Reservation
Separate screen for
reserving hotel for a
group
Reserve using
reservation screen
Individual
Reservation
Separate screen for
reserving hotel for an
individual
Reserve using
reservation screen
Allocating
Rooms
Two screens one
for the groups and
one for the
individuals, dynamic
search according to
check in and check
out date.
One allocating screen
shows the empty
rooms, not dynamic,
working parallel with
manual system
Search
Every screen has
search capability for a
customer or group
and its customers,
able to search by
name, room number,
date, group or name.
Search for customers
by their names or
room numbers.
Settings
Able to change screen
colors, hotel logo,
and hotel name.
"No settings"
Reports
Static reports on each
screen and a report
generator for
individuals and
groups.
Four static reports on
reservation screen.
Front Office
Manager
Provide a screen with
ability to add a
recipient, room,
facility, and charts
about the hotel status.
Screen for adding
new rooms and its
status.
Vol. 2, No. 10, 2011

52 | P a g e
countries . His research is concentrated on Automation in
different branches.

Vol. 2, No. 10, 2011

53 | P a g e
Automatic Classification and Segmentation of Brain
Tumor in CT Images using Optimal Dominant Gray
level Run length Texture Features

A.Padma
Research scholar
Thiyagarajar college of engg
Tamil nadu-6250015,INDIA
R.Sukanesh
Professor of ECE Dept
Thiyagarajar college of engg
Tamil nadu-6250015,INDIA

Abstract Tumor classification and segmentation from brain
computed tomography image data is an important but time
consuming task performed manually by medical experts.
Automating this process is challenging due to the high diversity
in appearance of tumor tissue among different patients and in
many cases, similarity between tumor and normal tissue. This
paper deals with an efficient segmentation algorithm for
extracting the brain tumors in computed tomography images
using Support Vector Machine classifier. The objective of this
work is to compare the dominant grey level run length feature
extraction method with wavelet based texture feature extraction
method and SGLDM method. A dominant gray level run
length texture feature set is derived from the region of interest
(ROI) of the image to be selected. The optimal texture features
are selected using Genetic Algorithm. The selected optimal run
length texture features are fed to the Support Vector Machine
classifier (SVM) to classify and segment the tumor from brain
CT images. The method is applied on real data of CT images of
120 images with normal and abnormal tumor images. The
results are compared with radiologist labeled ground truth.
Quantitative analysis between ground truth and segmented
tumor is presented in terms of classification accuracy. From the
analysis and performance measures like classification accuracy, it
is inferred that the brain tumor classification and segmentation
is best done using SVM with dominant run length feature
extraction method than SVM with wavelet based texture feature
extraction method and SVM with SGLDM method. In this
work,we have attempted to improve the computing efficiency as
it selects the most suitable feature extration method that can used
for classification and segmentation of brain tumor in CT images
efficiently and accurately. An avearage accuracy rate of above
97% was obtained usinh this classification and segmentation
algorithm.
Keywords - Dominant Gray Level Run Length Matrix method
(DGLRLM); Support Vector Machine (SVM); Spatial Gray Level
Dependence Matrix method (SGLDM); Genetic Algorithm(GA).

I. INTRODUCTION
In recent years, medical CT Images have been applied in
clinical diagnosis widely. That can assist physicians to detect
and locate Pathological changes with more accuracy.
Computed Tomography images can be distinguished for
different tissues according to their different gray levels. The
images, if processed appropriately can offer a wealth of
information which is significant to assist doctors in medical
diagnosis. A lot of research efforts have been directed towards
the field of medical image analysis with the aim to assist in
diagnosis and clinical studies [1]. Pathologies are clearly
identified using automated CAD system [2]. It also helps the
radiologist in analyzing the digital images to bring out the
possible outcomes of the diseases. The medical images are
obtained from different imaging systems such as MRI scan, CT
scan and Ultra sound B scan. The computerized tomography
has been found to be the most reliable method for early
detection of tumors because this modality is the mostly used in
radio therapy planning for two main reasons. The first reason is
that scanner images contain anatomical information which
offers the possibility to plan the direction and the entry points
of radio therapy rays which have to target only the tumor
region and to avoid other organs. The second reason is that CT
scan images are obtained using rays, which is same principle as
radio therapy. This is very important because the intensity of
radio therapy rays have been computed from the scanned
image.
Advantages of using CT include good detection of
calcification, hemorrhage and bony detail plus lower cost, short
imaging times and widespread availability. The situations
include patient who are too large for MRI scanner,
claustrophobic patients, patients with metallic or electrical
implant and patients unable to remain motionless for the
duration of the examination due to age, pain or medical
condition. For these reasons, this study aims to explore
methods for classifying and segmenting brain CT images.
Image segmentation is the process of partitioning a digital
image into set of pixels. Accurate, fast and reproducible image
segmentation techniques are required in various applications.
The results of the segmentation are significant for classification
and analysis purposes. The limitations for CT scanning of head
images are due to partial volume effects which affect the edges
produce low brain tissue contrast and yield different objects
within the same range of intensity. All these limitations have
made the segmentation more difficult. Therefore, the
challenges for automatic segmentation of the CT brain images
have many different approaches. The segmentation techniques
Vol. 2, No. 10, 2011

54 | P a g e
proposed by Nathali Richarda et al and Zhang et al [3][4]
include statistical pattern recognition techniques. Kaiping et al
[5] introduced the effective Particle Swarm optimization
algorithm to segment the brain images into Cerebro spinal fluid
(CSF) and suspicious abnormal regions but without the
annotation of the abnormal regions. Dubravko et al and
Matesin et al [6] [7] proposed the rule based approach to label
the abnormal regions such as calcification, hemorrhage and
stroke lesion. Ruthmann.et al [8] proposed to segment Cerebro
spinal fluid from computed tomography images using local
thresholding technique based on maximum entropy principle.
Luncaric et al proposed [9] to segment CT images into
background, skull, brain, ICH, calcifications by using a
combination of k means clustering and neural networks. Tong
et al proposed [10] to segment CT images into CSF, brain
matter and detection of abnormal regions using unsupervised
clustering of two stages. Clark et al [11] proposed to segment
the brain tumor automatically using knowledge based
techniques. Lauric et al [12] proposed to segment the CT brain
data into CSF and brain matter using Bayesian classifier.
Genesan et al introduced [13] to segment the tumor from CT
brain images using genetic algorithm. Hu et al proposed [14] to
segment the brain matter from 3D CT images by applying
Fuzzy C means and thresholding to 2D slices to create 2D
masks and then propagating 2D masks between neighboring
slices. From the above literature survey shows that intensity
based statistical features are the straightest forward and have
been widely used, but due to the complexity of the pathology in
human brain and the high quality required by clinical diagnosis,
only intensity features cannot achieve acceptable result. In such
applications, segmentation based on textural feature methods
gives more reliable results. Therefore texture based analysis
have been presented for tumor segmentation such as Dominant
gray level run length matrix method ,SGLDM method and
wavelet based texture features are used and achieve promising
results.
Based on the above literature, better classification accuracy
can be achieved using dominant run length statistical texture
features. In this paper, the authors would like to propose a
dominant gray level run length statistical texture feature
extraction method [15] The extracted texture features are
optimized by Genetic Algorithm(GA)[16] for improving the
classification accuracy and reducing the overall complexity.
The optimal texture features are fed to the SVM [17]
classifier to classify and segment the tumor region in brain
CT images.
II. MATERIALS AND METHODS
Most classification techniques offer grey level (i.e) pixel
based statistical features. The proposed system is divided into 4
phases (i) Image preprocessing ii) Segmentation of Region of
Interest (ROI) (iii) Feature extraction (iv) Feature selection (v)
Classification and Evaluation. For feature extraction, we
discovered three methods which are i) Dominant gray level
run length feature extraction method ii) Wavelet based feature
extraction method iii) SGLDM method . Once all the features
are extracted, then for feature selection, we use Genetic
Algorithm (GA) to select the optimal run length statistical
texture features. After selecting the optimal run length texture
features, to classify and segment the tumor region from brain
CT images using SVM classifier.
A. Image preprocessing
Brain CT images are noisy, inconsistent and incomplete,
thus preprocessing phase is needed to improve the image
quality and make the segmentation results more accurate. The
cropping operation can be performed to remove the
background. The Contrast Limited Adaptive Histogram
Equalization (CLAHE) can be used to enhance the contrast
within the soft tissues of the brain images and hybrid median
filtering technique can also be used to improve the image
quality.
B. Selection of ROI
First the pixel having highest intensity value in the image is
selected, then that pixel is compared to the neighboring pixels.
The comparison goes till there is change in the intensity level
of pixel value. All the pixels having the similar intensity form
the Region of Interests.
C. Feature extraction
As the tissues present in brain are difficult to classify using
shape or intensity level of information, the texture feature
extraction is founded to be very important for further
classification. The purpose of feature extraction is to reduce
original data set by measuring certain features that distinguish
one region of interest from another .The spatial features from
each ROI are extracted by dominant gray level run length
method. In this method, by using multilevel domonant
eigenvector estimation algorithm and the Bhattacharyya
distance measure for texture extraction.
D. Dominant Gray Level Run Length Matrix method
The DGLRLM is based on computing the number of gray
level runs of various lengths. A gray level run is a set of
consequence and collinear pixels point having the same gray
level value. The length of the run is the number of pixel points
in the run. The gray level run length matrix is as follows:
| (d, ) = [p(i, j / d, )], 0 < i s Ng, 0 < i s Rmax
(1)
Where Ng is the maximum gray level and Rmax is the
maximum Run length. The element p (i, j | ) specified the
estimated number of runs that a given image contains a run
length j for a gray level i in the direction of angle . Four gray
level run length matrices corresponding to = 0o, 45o, 90o,
135o are computed for an each region of interest and the
following four textural features such as Short run low gray
level emphasis, short run high gray level emphasis, Long run
low gray level emphasis, Long run high gray level emphasis
are extracted for each gray level run length matrix and take the
average all four gray level run length matrices.
1.Short Run Low Gray Level Emphasis(SRLGE)
SRLGE = 2 2
1 1
.
) , (
/ 1
j i
j i p
nr
N
j
M
i

= =
(2)
2. Short Run High Gray Level Emphasis(SRHGE)
Vol. 2, No. 10, 2011

55 | P a g e
SRLGE = 2
2
1 1
. ). , (
/ 1
j
i j i p
nr
N
j
M
i

= =
(3)
3. Long Run Low Gray Level Emphasis(LRLGE)
LRLGE = 2
2
1 1
). , (
/ 1
i
j j i p
nr
N
j
M
i

= =

(4)
4. Long Run High Gray Level Emphasis(LRHGE)
LRHGE =

= =
N
j
M
i
j i j i p nr
1
2 2
1
. ). , ( / 1
(5)
Where p is the run length matrix, p (i, j) is an element of the
run length matrix at the position (i, j) and nr is the number of
runs in the image. For all the four directions, the dominant
gray level run length texture features are extracted.
TABLE 1. FEATURES EXTRACTED USING DGLRLM METHOD
SI-NO High order DGLRLM features
1. Short Run Low Gray Level emphasis(SRLGE)
2. Short Run High Gray Level emphasis(SRHGE)
3. Long Run Low Gray Level emphasis(LRLGE)
4. Long Run High Gray level emphasis(LRLGE)

a) Wavelet based feature extraction method
A two level discrete wavelet decomposition [18] of region
of interest (ROI) is performed which results in four sub bands
such as LL,LH,HH,HL. The sub band LL represents low
frequency pat of the image and the sub bands LH,HH,HL
represents high frequency part of the image. The wavelet
decomposition transform is applied this high frequency sub
bands only. Daubechies wavelet filter of order two is used.
Algorithm for feature extraction is as follows
- Obtain the sub-image blocks, starting from the top
left corner.
- Decompose sub-image blocks using two level 2-D
DWT.
- Derive Spatial Gray Level Dependence Matrices
(SGLDM) or Gray Level Co-occurrence matrices
[19] for each 2 level high frequency sub-bands of
decomposed sub image blocks with 1 for distance
and 0, 45, 90 and 135 degrees for and averaged.
From these co-occurrence matrices, the following nine
Haralick second order statistical texture features [20] called
wavelet Co-occurrence Texture features (WCT) are extracted .
TABLE 2. WCT FEATURES EXTRACTED USING SGLDM METHOD
SI.No Second order WCT features

1. Entropy (Measure the disorder of an image)
2. Energy ( Measure the textural uniformity
3. Contrast (Measure the local contrast in an image)
4. Sum Average ( average of the gray levels)
5. Variance ( heterogeneity of an image)
6. Correlation (correlation of pixel pairs)
7.Maxprobability(the most prominent pixel pair)
8. Inverse Difference moment( homogeneity )
9. Cluster tendency (Measure the grouping of
Pixels with similar values)
b) Spatial Gray Level Dependence Matrix method
(SGLDM)
For each ROI, the Spatial Gray Level Dependence Matrix
method (SGLDM) can be used to extract the second order
statistical texture features for the better diagnosis. This method
is based on the estimation of the second order joint conditional
probability density functions P(i,j | d,u) where u = 0,45,90,135
degrees. Each P (i, j |d,u) is the probability matrix of two pixels
which are located with an inter sample distance d and direction
u have a gray level i to gray level j. The estimated value for
these probability density functions can be represented by
| (d, ) = [P (i, j / d, )], 0 < i, j s Ng
(6)
Where, Ng is the maximum gray level. In this method,
four gray level co-occurrence matrixes for four different
directions ( = 0o, 45o, 90o, 135o) are obtained for a given
distance d (=1, 2) and the following 13 Haralick statistical
texture features [20] are calculated for each gray level co-
occurrence matrix and take the average of all four gray level
co-occurrence matrices. The set of extracted second order
statistical texture features from each ROI forms the feature
vector or feature set.
TABLE 3.FEATURES EXTRACTED USING SGLDM METHOD
SI-no Second order SGLDM features
1. Energy
2. Entropy
3. Contrast
4. Inverse difference moment
5. Homogeneity
6. Correlation
7. Sum Average
8. Sum Entropy
9. Difference Average
10. Difference Entropy
11. Sum of squares variance
12. Information measures of
Correlation I& II
c) Feature selection
Feature selection is the process of choosing subset of
features relevant to particular application and improves
classification by searching for the best feature subset, from the
fixed set of original features according to a given feature
evaluation criterion(ie., classification accuracy). Optimized
feature selection reduces data dimensionalities and
computational time and increase the classification accuracy.
The feature selection problem involves the selection of a subset
of the features from a total number of the features, based on a
given optimization criterion. T denotes the subset of selected
features and V denotes the set of remaining features. So, S = T
Vol. 2, No. 10, 2011

56 | P a g e
U V at any time. J(T) denotes a function evaluating the
performance of T. J depends on the particular application. Here
J(T) denotes the classification performance of classifying and
segmenting soft tissues from brain CT images using the set of
features in T. In this work, genetic algorithm is used.
GENETIC ALGORITHM:
We consider the standard GA to begin by randomly
creating its initial population. Solutions are combined via a
crossover operator to produce offspring, thus expanding the
current population of solutions. The individuals in the
population are then evaluated via a fitness function, and the
less fit individuals are eliminated to return the population to its
original size. The process of crossover, evaluation, and
selection is repeated for a predetermined number of generations
or until a satisfactory solution has been found. A mutation
operator is generally applied to each generation in order to
increase variation. In the feature selection formulation of the
genetic algorithm, individuals are composed of chromosomes:
a 1 in the bit position indicates that feature should be selected;
0 indicates this feature should not be selected. As an example
the chromosome 00101000 means the 3rd and 5th features are
selected. The features 1, 2, 4, 6, 7, 8th feature are not selected.
That is the chromosome represents T={3,5} and
V={1,2,4,6,7,8}. Fitness function for the given chromosome
T is defined as
Fitness (T) = J(T) - penalty(T) (7)
where T is the corresponding feature subset , and
penalty(T) = w * (|T| -d) with a penalty coefficient w. The size
value d is taken as a constraint and a penalty is imposed on
chromosomes breaking this constraint. The chromosome
selection for the next generation is done on the basis of fitness.
The fitness value decides whether the chromosome is good or
bad in a population. The selection mechanism should ensure
that fitter chromosomes have a higher probability survival. So,
the design adopts the rank-based roulette-wheel selection
scheme. If the mutated chromosome is superior to both parents,
it replaces the similar parent. If it is in between the two parents,
it replaces the inferior parent; otherwise, the most inferior
chromosome in the population is replaced. The selected
optimal feature set based on the test data set is used to train the
SVM classifier to classify and segment the tumor region from
brain CT images.
a) Classifier
Classification is the process where a given test sample is
assigned a class on the basis of knowledge gained by the
classifier during training. To make the classification results
comparable and for exhaustive data analysis, we have used
leave one out classification method for the SVM classifier.
Figure 1 shows the class separation of SVM classifier.
Support Vector Machine (SVM) performs the robust non-linear
classification with kernel trick. SVM is independent of the
dimensionality of the feature space and that the results obtained
are very accurate. It outperforms other classifiers even with
small numbers of available training samples. SVM is a
supervised learning method and is used for one class and n
class classification problems.

Figure 1.SVM classifier
It combines linear algorithms with linear or non-linear
kernel functions that make it a powerful tool in the machine
learning community with applications such as data mining and
medical imaging applications. To apply SVM into nonlinear
data distributions, the data can be implicitly transformed to a
high dimensional feature space where a separation might
become possible. For a binary classification given a set of
separable data set with N samples X = {Xi}, i = 1, 2 . N,
labeled as Yi = 1. It may be difficult to separate these 2
classes in the input space directly. Thus they are mapped into a
higher dimensional feature space by X = f(x).
The decision function can be expressed as
f(x) = W.x + (8)
Where W.x + P = 0 is a set of hyper planes to separate the
two classes in the new feature space. Therefore for all the
correctly classified data,
Yi f(x) = Yi (W.x + ) > 0, i = 1, 2 .. N (9)
By scaling W and properly, we can have f(x) = W.x +
= 1 for those data labeled as +1 closes to the optimal hyper
plane and f(x) = W.x + = -1 for all the data labeled as -1
closes to the optimal hyper plane. In order to maximize the
margin the following problem needs to be solved.
Min (||W||2/2)
Subject to Yi f(x) = Yi (W.x + ) 1, i = 1, 2 .. N (10)
It is a quadratic programming problem to maximize the
margins which can be solved by sequential minimization
optimization. After optimization, the optimal separating hyper
plane can be expressed as
f(x) =
=
+
N
i
i i i
x x K Y
1
) , ( o
(11)
Where K(.) is a kernel function, is a bias, o is the
solutions of the quadratic programming problem to find
maximum margin. When o is non-zero, are called support
vectors, which are either on or near separating hyper plane. The
decision boundary (i.e.) the separating hyper plane whose
decision values f(x) approach zero, compared with the support
vectors, the decision values of positive samples have larger
positive values and those of negative samples have larger
negative values.
Vol. 2, No. 10, 2011

57 | P a g e
Therefore the magnitude of the decision value can also be
regarded as the confidence of the classifier. The larger the
magnitude of f(x), the more confidence of the classification by
choosing a Gaussian kernel function
K(x,y) = e
-
||x-y||
2
(12)
Where the value of was chosen to be 1 and has good
performance for the following two reasons. First reason is the
Gaussian model has only one parameter and it is easy to
construct the Gaussian SVM classifier compared to polynomial
model which has multiple parameters. Second reason is there is
less limitation in using Gaussian kernel function due to
nonlinear mapping in higher dimensional space.
III. RESULTS AND DISCUSSIONS
This section describes the comparative study of
classification performance of the SVM classifier for different
texture analysis methods used for classification and
segmentation of tumor from brain CT images. The texture
features extracted from each ROI of the image to be selected by
using pixel based intensity method. The texture features are
extracted with same set of images and are obtained from 16 bit
gray level images. The SVM is used as a classifier. The
results from the SVM classifier for all the texture analysis
methods are evaluated by using the statistical analysis.
An experiment has been conducted on a real CT scan brain
images based on the proposed system.

(a) CT image (b) Hist-Equalized image \

c) HMF Image
Figure 2(a-c) Preprocessed image
Figure 2(a-c) shows the original input CT image, histogram
equalized image, hybrid median filtered image which is used to
reduce the different illumination conditions and noises.

Figure 3(a-d) Selection of ROIS
Figure 3(a-d) represents the selected ROIs of the image to
be segmented using pixel based intensity method. For the
comparative analysis of texture analysis methods using SVM
classifier, 120 images were partitioned arbitrarily into training
set, testing set with equal number of images. The accuracy of
the classifier for the texture analysis methods are evaluated
based on the error rate. This error rate can be described by the
terms true and false positive and true and false negative as
follows
True Positive (TP): Abnormal cases correctly classified
True Negative (TN): Normal cases correctly classified.
False Positive (FP): Normal cases classified abnormal
False Negative (FN): Abnormal cases classified normal
The above terms are used to describe the clinical efficiency
of the classification and segmentation algorithm.
Sensitivity = TP / (TP + FN) * 100
Specificity = TN / (TN + TP) * 100
Accuracy = (TP + TN) / (TP + TN + FP + FN) * 100.
Accuracy is the proportion of correctly diagnosed cases
from the total number of cases. Sensitivity measures the ability
of the method to identify abnormal cases. Specificity measures
the ability of the method to identify normal cases. To make the
classification results comparable and for exhaustive data
analysis, leave one out cross validation method can be used to
estimate the classifier performance in unbiased manner. Here
each step, one data set is left out and the classifier is trained
using the rest and the classifier is applied to the left out data
set. This procedure is repeated such that each data set is left out
once. In our application for evaluating classification accuracy,
10 fold cross validation method is done on the data set
collected from 120 images (60 normal and 60 abnormal).
The images are divided into 10 sets each consisting of 6
normal images and 6 abnormal images. Then 9 sets are used for
training and remaining set is used for testing. In next iteration
(2-10), 9 sets are used for training and remaining set is used for
testing. This process is repeated for 10 times. Hence 10
iterations are done. The classification accuracy was calculated
for by taking the average of all correct classifications. All the
textural features were normalized by the sample mean and
standard deviation of the image.

Vol. 2, No. 10, 2011

58 | P a g e
TABLE 4. CLASSIFICATION RESULTS OF SVM CLASSIFIER USING CO-
OCCURRENCE, WCT, NEW RUN LENGTH FEATURES
Feature feature features classification
name length selected accuracy

1.Co-occu 52 16 93.3%
rrence
2.2 level-WCT
features 27 12 95.8%
3.New run length
features 16 8 98.3%

Figure 4. No of detections based on different texture features
Figure 4 shows the number of abnormal tumor detection
based on the three texture analysis methods. The results show
that, if the representative samples increased, it gives good
classification accuracy for 10 fold cross validation method.
The sensitivity and specificity values of texture analysis
methods are shown in Figure 5.

Figure 5. Sensitivity,specificity of texture anlysis methods
The sensitivity, specificity values of the new run length
method are 96.72%, 98.3% and the wavelet co-occurrence
method are 96.6%,95% and co-occurrence method are 95%,
91.6% respectively.
Then the accuracy of the new run length method is 98.3%
and wavelet co-occurrence method is 95.8% and the co-
occurrence method is 93.3% respectively. We now compare the
new run length method with the co-occurrence method and
wavelet co-occurrence method on the brain CT images. For the
co-occurrence method, 13 co-occurrence features are
computed for each of the four directions (i.e) 0,45,90,135
degrees; for the wavelet co-occurrence method, the texture
features used for each wavelet decomposition high frequency
sub bands are entropy, energy, contrast , sum average, variance,
correlation, max probability, inverse difference moment,
cluster tendency. Table 5 shows the accuracy of different
segmentation methods.
TABLE 5.ACCURACY OF DIFFERENT SEGMENTATION METHODS
SI-n Segmentation method Accuracy
1. Fuzzy C means 85%
2. K means 87.3%
3. Bayesian classifier 89%
4. Genetic algorithm 93%
5. SVM classifier with
feature extraction method 98.3%
The same genetic feature selection method is used for all
the three texture analysis methods. In dominant gray level run
length feature extraction method, the selected features were
long run high gray level emphasis and long run low gray level
emphasis. The long run high gray level emphasis captures the
inhomogeneous nature of texture features and long run low
gray level emphasis captures the homogeneous nature of
texture features. About 93% classification accuracy is achieved
by most of the feature vectors.

Figure 6(a,b,c,d). CT image and the segmented tumor image
Figure 6 (a,b,c,d) shows the input CT image and the
corresponding segmented tumor image. From the classification
results of the SVM classifier for all the three texture analysis
methods, the new run length method perform comparably well
with the wavelet co-occurrence features and better than the co-
occurrence features. This demonstrates that there is more
texture information contained in run length matrices and that a
good method of extracting such information is important to
classification and segmentation algorithm.
0
20
40
60
80
NEW RUN LENGTH 2 level WCTCo-occurrence
N
U
M
B
E
R

O
F

D
E
T
E
C
T
I
O
N
S

Different texture features
Results of segmentation of soft tissues based on
different texture features
TP
TN
FP
FN
86%
88%
90%
92%
94%
96%
98%
100%
102%
NEW RUN
LENGTH
2 level WCT Co-
occurrence Different texture features
senstivity,specificity of different texture features
sensitivity
specificity
Vol. 2, No. 10, 2011

59 | P a g e
IV. CONCLUSION
In this work a new dominant run length feature extraction
methodology for the classification and segmentation of tumor
in brain CT images using support vector machine with genetic
algorithm feature selection is proposed. The algorithm has been
designed based on the concept of different types of brain soft
tissues (CSF, WM, GM, Abnormal tumor region) have
different textural features. The selection method of ROI is
simple and accurate. The results show that the segmentation
and classification of tumor for the new run length feature
extraction method yields better results compared to the other
texture analysis methods based on SVM classifier. It is found
that this method gives favorable result with accuracy
percentage of above 98% for the images that are being
considered. This would be highly useful as a diagnostic tool for
radiologists in the automated segmentation of tumor in CT
images.
The goal of this work is to compare the classification
performance of the SVM classifier based on different texture
analysis methods. Hence it is concluded that the neural network
supported by conventional texture analysis methods can be
effectively used for classification and segmentation of brain
tumor from CT images. Use of large data bases is expected to
improve the system robustness and ensure the repeatability of
the resulted performance. The automation procedure proposed
in this work using a SVM enables proper abnormal tumor
region detection and segmentation thereby saving time and
reducing the complexity involved.
In this research work, the four segmentation methods,
Bayesian classification, fuzzy C means, K means and
expectation maximization algorithm were produced good
results. In this work, the dominant feature extraction method
with SVM classifier is proposed for the classification and
segmentation of brain CT images and the results are compared
with Bayesian classification and fuzzy c means, k means
clustering method and expectation maximization algorithm
through statistical analysis. The proposed method has better
performance compared to the other existing methods. Plans for
future work include the specific annotation of the abnormal
regions such as hemorrhage, calcification and lesion.
REFERENCES
[1]. Duncan J.S.,Ayache N., Medical Image Analysis- Progress Over
two decade and challenges ahead, IEEE Trans on PAMI , Vol 22,pp. 85
106,2000.
[2]. G.P.Tourassi,, Journey towards computer aided Diagnosis Role of
Image Texture Analysis, Radiology, Vol 2, pp.317 320,1999.
[3]. Nathalie Richard, Michel Dojata, Catherine Garbayvol., Distributed
Markovian Segmentation Application to MR brain Scans, Journal of
Pattern Recognition, Vol 40, pp.3467 3480,2007.
[4]. Y. Zhang, M. Brady, S. Smith, Segmentation of Brain MR Images
through hidden Markov random field model and the expectation-
maximization algorithm, IEEE Transactions on Medical Imaging, Vol
20, pp. 45 57,2001.
[5]. Kaiping Wei, Bin He,Tao Zhang, Xianjun Shen, A Novel Method for
segmentation of CT Head Images, International conference on Bio
informatics and Biomedical Engineering, Vol 4 , pp. 717 720, 2007.
[6]. Dubravko Cosic ,Sven Loncaric, Rule based labeling of CT head
images, 6th conference on Artificial Intelligence in Medicine, pp. 453
456,1997.
[7]. Matesn Milan, Loncaric Sven, Petravic Damir, A rule based approach to
stroke lesion analysis from CT brain Images, 2nd International
symposium on Image and Signal Processing and Analysis, June (2001),
pp. 219 223,2001.
[8]. Ruthmann V.E., Jayce E.M., Reo D.E, Eckaidt M.J.,(1993), Fully
automated segmentation of cerebro spinal fluid in computed
tomography, Psychiatry research: Neuro imaging, Vol 50 , pp.101
119,1993.
[9]. Loncaric S and D. Kova Cevic, A method for segmentation of CT head
images, Lecture Notes on Computer Science, Vol 1311,pp. 1388
305,1993.
[10]. Tong Hau Lee, Mohammad Faizal, Ahmad Fauzi and Ryoichi Komiya,
Segmentation of CT Brain Images using unsupervised clusterings,
Journal of Visualization, Vol 12, pp.31-138,2009.
[11]. Clark M C., Hall L O., Goldgof D B, Velthuzien R., Murtagh F R., and
Silbiger M S., Automatic tumor segmentation using knowledge based
techniques, IEEE Transactions on Medical Imaging, Vol 17, pp. 187-
192,1998.
[12]. Lauric A.,Frisken S., Soft segmentation of CT brain data, Technical
Report TR-2007-3 Tuffs University, M A.
[13]. Ganesan R, Radhakrishnan R., Segmentation of Computed Tomography
Brain Images using genetic algorithm, International Journal of Soft
computing, Vol .4,pp.157-161,2009.
[14]. Hu Q., Qian G, Aziz A and Nowinski W L., Segmentation of brain
from computed tomography head images, IEEE Eng. Medicine and
Biology Society , Vol 6, pp.3375-3378,2005.
[15]. Tang Xiaoou, Texture Information in Run Length Matrices, IEEE
Transaction on Image Processing, Vol 7, pp.234-243, 1998.
[16]. Frank Z. Brill., Donald E. Brown., and Worthy N. Martin, Fast Genetic
Selection of Features for Neural Network Classifiers, IEEE Trans.,on
Neural Networks ,Vol 3, pp. 324-328,1992.
[17]. El-Naqa I, Yang Y, Wernick M N, Galatsanos N P and Nishikawa R M,
A support vector machine approach for detection of micro
calcifications, IEEE Trans. on Medical Imaging, Vol 21, pp.1552-
1563,2002.
[18]. Van G., Wouver P.,Scheunders and D.Van Dyck, Statistical texture
characterization from discrete wavelet representation, IEEE Trans.
Image processing ,Vol 8, pp.592-598,1999.
[19]. Haddon J F, Boyce J F, Co-occurrence Matrices for Image analysis. IEE
Electronic and Communications Engineering Journal , Vol 5,pp. 71
83,1993.
[20]. Haralick R M, Shanmugam K and Dinstein I, Texture features for
Image classification, IEEE Transaction on System, Man, Cybernetics,
Vol 3, pp. 610 621,1973.
[21]. Chaabane, L. (2011). Improvement of Brain Tissue Segmentation Using
Information Fusion Approach. IJACSA - International Journal of
Advanced Computer Science and Applications, 2(6), 84-90.
[22]. Chaabane, L. (2011). Evaluation of the Segmentation by Multispectral
Fusion Approach with Adaptive Operators: Application to Medical
Images. IJACSA - International Journal of Advanced Computer Science
and Applications, 2(9), 1-7.
Vol. 2, No. 10, 2011

60 | P a g e
An Ontology- and Constraint-based Approach for
Dynamic Personalized Planning in Renal Disease
Management

Normadiah Mahiddin
1
, Yu-N Cheah
2
, Fazilah Haron
3

1
Faculty of Information Science and Technology, Universiti Kebangsaan Malaysia, 43600 Bangi, Malaysia

2, 3
School of Computer Sciences, Universiti Sains Malaysia, 11800 USM Penang, Malaysia
3
College of Computer Science and Engineering, Taibah University, P.O. Box 30002, Madinah, Saudi Arabia

AbstractHealthcare service providers, including those involved
in renal disease management, are concerned about the planning
of their patients treatments. With efforts to automate the
planning process, shortcomings are apparent due to the following
reasons: (1) current plan representations or ontologies are too
fine grained, and (2) current planning systems are often static. To
address these issues, we introduce a planning system called
Dynamic Personalized Planner (DP Planner) which consists of:
(1) a suitably light-weight and generic plan representation, and
(2) a constraint-based dynamic planning engine. The plan
representation is based on existing plan ontologies, and developed
in XML. With the available plans, the planning engine focuses on
personalizing pre-existing (or generic) plans that can be
dynamically changed as the condition of the patient changes over
time. To illustrate our dynamic personalized planning approach,
we present an example in renal disease management. In a
comparative study, we observed that the resulting DP Planner
possesses features that rival that of other planning systems, in
particular that of Asgaard and O-Plan.
Keywords-patient care planning; treatment protocols; dynamic
treatment planning; personal health services.
I. INTRODUCTION
Healthcare service providers are undoubtedly concerned
about updating their patients health records or profiles, and the
planning of their patients treatments to support the efficient
and effective delivery of healthcare services. However, not all
healthcare service providers are carrying out planning activities
effectively, especially when it comes to automated or
computer-based planning, due to shortcomings in current
planning systems.
The first problem is that most of the current plan
representations or ontologies are too fine grained (detailed).
This means that the plan representations or ontologies are not
suited for all situations and for all levels. We need to have a
portable and intuitively easy representation that facilitates the
storage and manipulation of generic plans. The second problem
is that current planning systems are often static. This means
planning is carried out once without taking into account
changes that may take place as time goes on. These plans also
do not consider past events. Dynamic planning is therefore
required to allow plans to be updated as new situations arise.
To address the concerns above, we present a methodology
for generic and dynamic healthcare planning, resulting in a
system called the Dynamic Personalized Planner (DP Planner)
[1]. For this purpose, we define (1) a suitably light-weight and
generic plan representation based on existing plan ontologies,
and (2) a constraint-based dynamic planning engine.
II. STATE OF THE ART
A popular approach for plan representation is via
ontologies, i.e. plans are designed based on project specific
ontologies and domain description languages [2]. Fig. 1 shows
the structure of the Plan Ontology proposed by Tate [3].

Figure 1. Plan Ontology structure [3].
Another useful plan ontology for generic planning is the
task-specific ontology and approach called Asbru which uses
skeletal plans [4]. Asbru represents skeletal plans using a task-
specific, intention-based, and time-oriented language. It can be
used to design specific plans [5]. In Asbru, a plan contains a
name and a set of arguments (time annotation and knowledge
roles). All plans and actions have a temporal dimension and the
plans execution is controlled by a number of conditions such
as filter, setup, suspend, reactivate, abort and complete [6]. Fig.
2 shows the Asbru plan ontology structure.
Besides plan ontologies (which contribute towards plan
representation), there are also a number of plan generation
initiatives. An example of such an initiative is the RAX
Planner/Scheduler (RAX-PS) [7]. The RAX-PS generates plans
Vol. 2, No. 10, 2011

61 | P a g e
that are temporally flexible, allowing the execution system to
adjust to actual plan execution conditions without breaking the
plan. The result is a system capable of building concurrent
plans.

Figure 2. The Asbru plan ontology structure [4].
Fig. 3 shows the architecture of the planning system in
RAX-PS. For the RAX-PS, the domain model describes the
dynamics of the system [8] to which the planner is being
applied, i.e. the Deep Space One Spacecraft. The plan database
is initialized by a plan request which consists of an initial state
and a set of goals. The initialized database is then modified by
a search engine to generate a complete and valid plan. This
complete plan is then put into operation by an execution agent.
The heuristics and planning experts are also integral parts of
the Deep Space One planning system. The heuristics guide the
search engine while the planning expert interfaces with external
systems which provide critical inputs such as altitude and
speed.

Figure 3. The RAX Planner/Scheduler (RAX-PS) [7].
The Capture the Flag (CTF) [9] game project uses the
notion of critical points (time during the execution of an action
or plan where a decision might be made) to define states in the
continuous domain. These states are then used to efficiently
evaluate plans. An action or a plan posts a goal, G. This
invokes the CTF planning algorithm [10].
In the effort to generate outputs that are dynamically
assembled from smaller fragments, the Personalized Healthcare
Information (PHI) system [11] composes personalized
documents that conform to an individuals health profile. The
composition of PHI is carried out in a three-step procedure
which are (1) selection of a set of Topic Specific Documents
(TD), where each selected TD addresses some of the
individuals healthcare concerns, (2) combination of the
selected TDs to produce an aggregated PHI document, and (3)
verification of the accuracy of the aggregated PHI document.
Each individual illness/concern/issue noted in the health profile
is addressed by at least a single TD. Constraint satisfaction
techniques are used to ensure that the aggregated PHI
document is medically consistent. Fig. 4 shows the processes
for PHI composition.

Figure 4. The process flow for PHI composition [11].
We have found the PHI systems approach in using
constraints attractive as this approach could be adapted in our
DP Planner to ensure the coherency of plan fragments that are
assembled to form a complete plan.
III. THE DYNAMIC PERSONALISED PLANNING
METHODOLOGY
The dynamic personalized planning approach consists of
two phases:
1. Plan ontology definition and representation.
2. Planning algorithm definition.
Fig. 5 briefly shows the two phases together with the
techniques and approaches used.

Figure 5. DP Planner methodology with related techniques and
approaches [1].
Vol. 2, No. 10, 2011

62 | P a g e
A. Phase 1: Plan Ontology Definition and Representation
In defining the plan ontology, the Asbru plan ontology was
adopted as the basis for the DP Planners plan ontology in view
that it is reasonably concise compared to other ontologies that
were surveyed. Besides this, some elements of Goals,
Operators, Methods and Selection (GOMS) analysis [12] were
also incorporated. GOMS is a method for analyzing and
modeling the knowledge and skills that a user must develop in
order to perform tasks on a device or system.
1) Plan ontology definition
Fig. 6 shows the DP Planners plan ontology [13].

Figure 6. DP Planners plan ontology structure [13].
Referring to Fig. 6, the plan is positioned at the highest
position in the plan hierarchy, and is basically the task that
needs to be performed. Examples of plans are those for kidney
patient treatment monitoring, gestational diabetes mellitus
monitoring, student performance monitoring, etc. The plan
consists of a sequence of plan fragments. These plan fragments
are the necessary steps to achieve the task and can be viewed as
the most crucial component of the plan ontology.
Each plan fragment consists of seven attributes:
Name: Identifies a plan.
Goal: States the target to be achieved.
Date: States the date of plan execution (if required).
Time: States the duration of plan execution.
Constraints: Information of the plan execution limit.
Condition: Situations in which the task takes place.
Status: Keeps track of the situation of plan execution.
Some of the plan components attributes have detailed sub-
attributes. Examples are as follows:
Date has six sub-attributes: Earliest Date Start (EDS),
Latest Date Start (LDS), Earliest Date Over (EDO),
Latest Date Over (LDO), Minimum Date Duration
(MinDD) and Maximum Date Duration (MaxDD),
Time also has six sub-attributes: Earliest Time Start
(ETS), Latest Time Start (LTS), Earliest Time Over
(ETO), Latest Time Over (LTO), Minimum Time
Duration (MinTD) and Maximum Time Duration
(MaxTD).
Condition consists of three sub-attributes:
o Pre-condition: This is to ensure that certain
conditions are met before the execution of a
particular plan fragment.
o Current condition: This ensures that certain
conditions are currently met in a particular
plan fragment.
o Post-condition: This is to ensure that certain
conditions are met after the execution of a
particular plan fragment and before the next
plan fragment.
2) Plan representation
Fig. 7 illustrates an example of a plan represented in XML
for the treatment of a patient with renal disease. Note that the
plan ontology can be naturally implemented in XML as the
hierarchical nature of ontologies maps very well into the nested
nature of XML. With the defined DP Planner plan ontology
and representation, the DP Planner planning engine
implementation is discussed next.

Figure 7. Plan representation in XML format.
B. Phase 2: Planning Algorithm Definition
The DP Planners planning strategy involves reusing plans
that are stored in the plan repository and subsequently
personalizing these plans according to the constraints defined
by the user.
The planning algorithm is divided into two parts (see Fig.
8):
1. Generic plan generation.
2. Plan personalization.

Vol. 2, No. 10, 2011

63 | P a g e

Figure 8. The planning algorithm.
1) Generic plan generation
Firstly, user inputs are matched against each of the existing
plans in the plan repository. A concentration unit is calculated
during the matching process to compare the closeness of the
match. This unit is based on the number of matches between
the users inputs (pre-conditions and current conditions) with
those of the plan fragments which constitute each plan. The
similarity between the users pre-condition and current
condition inputs, and those of a particular plan, P, is expressed
as C
P
, (see Equation 1).

(1)
where n = number of matches of pre-conditions + matches of
current conditions, and T
f
= total number of plan fragments in a
plan. The values of C for each plan will be compared. The plan
with the highest value will be chosen as the generic plan. After
a generic plan has been identified, the process of plan
personalization will follow
2) Plan personalization
The personalization method employs a combination of (1) a
simple matching and linking technique, and (2) a constraint-
based approach to form certain restrictions [9] so that only plan
fragments that meet the predetermined criteria and users inputs
can form the finalized and personalized plan for a particular
user or situation. Personalization is only carried out when the
users status is active, i.e. the plan is still relevant to the user's
condition. For a start, the user will be asked for the outcomes of
following the activities defined in a particular plan fragment.
These outcomes are called post-conditions. Here, a simple
matching technique will be applied to match the post-
conditions of a particular (current) plan fragment with the pre-
conditions in the next plan fragment. If these match each other,
that next plan fragment in the generic plan will be
recommended as the subsequent plan for the user. However, if
they do not match, the plan (or sequence of plan fragments)
will terminate at that current plan fragment.
The matching technique is applied to ensure that each plan
fragment in the finalized plan links to each other (see Fig. 9).
This is to ensure that the final plan generated by the system
corresponds to the needs of the user. After ensuring that the
plan fragments in the generic plan fulfils the initial matching of
post-conditions with the pre-conditions, the actual
personalization can then take place using constraints.

Figure 9. The linking between plan fragments in a plan.
Constraints are utilized to ensure the consistency of the
multiple fragments of a plan in order to form a complete plan.
The individual plan fragments must not contradict each other or
lead to improper recommendations [11]. Therefore, for the
purpose of the DP Planner, a constraint is simply a variable
which restricts the execution of a particular plan fragment. It
also describes the relationship between plan fragments in a
plan. Ultimately, the constraints are used to find suitable
replacements for plan fragments which are not suited for the
plan fragments preceding or following it. Fig. 10 illustrates
how constraints are used to personalize a generic plan.

Figure 10. The use of constraints in plan personalization.
Vol. 2, No. 10, 2011

64 | P a g e
In Fig. 10, the generic plans Plan Fragment 4 shows
constraint 5 (C5) = yes. However, let us assume this does not
fulfill a constraint specified by the user, i.e. the user has C5 =
no. Therefore, plan personalization is carried out to find a plan
fragment in the plan repository which fulfils the users
constraint. The example in Fig. 10 shows Plan Fragment 5 in
Plan A has C5 = no. Therefore, our system will use this plan
fragment to replace Plan Fragment 4 in the generic plan
provided other details are also met (e.g. pre-conditions, post-
conditions, etc.).
IV. RESULTS: EXAMPLE CASES
Fig. 11 shows the command-line system interface which
obtains inputs from the user to generate a generic and
personalized plan in the domain of renal disease.

Figure 11. The process of finding the best-matched plan in the plan
repository.
A. Generic Plan Generation
From Fig. 11, the interaction labeled A gets the inputs of
pre-conditions while the interaction labeled B gets the inputs of
current conditions from the user (bold text indicates text
entered by the user). Both of these conditions will be used to
match the pre-conditions and current conditions in the plans
found in the plan repository. As described in Section III.B.1,
this matching is carried out to find the plan that best matches
the users pre-condition and current condition inputs, i.e. the
plan with the highest concentration of matches value (see
Equation 1) is identified. Then, a copy of that plan is created as
an XML file and assigned with a new ID. The values for the
pre-conditions and current conditions are assigned with those
inputted by the user while dates, times, and post-conditions are
left empty.
B. Plan Personalization
After the generic plan has been generated, personalization
then takes place. Firstly, the user will be presented with the
first plan fragment in the plan as shown in Fig. 11 (labeled C).
Then, the planning system will prompt the user about their
status, i.e. whether active (user is implementing the plan) or not
active (not implementing the plan). Either response will result
in a different output for the user. Hence, we show the results of
different cases in the following sub-sections. This process is
repeated with other plan fragments (labeled D to H in Fig. 11).
1) Case 1: Status is active
This case is for situations when the user is implementing a
particular plan fragment. When this happens, the system will
prompt for details about the users post-condition as shown in
Fig. 12. Assuming that the users input for post-condition was
Percentage of renal damage = 95% therefore, this input will be
matched with the next plan fragments pre-condition. If they
match, the constraints of the next plan fragment will be
highlighted to the user: Diabetes = yes?: (T=True/F=False) as
shown in Fig. 12. If the user fulfils the constraint, the next plan
fragment will be presented. The output for this user will be
generated as shown in Fig. 13.

Figure 12. Part of the system interface when status of the user is active.

Figure 13. Part of the system output when status of the user is active.
2) Case 2: Status is not active
This case shows that the user is not implementing a
particular plan fragment. When this is indicated by the user, the
system will not perform any personalization in the subsequent
plan fragments. Therefore, the planning process is deemed to
Vol. 2, No. 10, 2011

65 | P a g e
be complete. The output for the user will be generated as
shown in Fig. 14. Here, the result is personalized by removing
the subsequent plan fragments that are not necessary.

Figure 14. Part of the system output when status of the user is not active.
3) Case 3: Inputs do not fulfill conditions or constraints in
the generic plan
This situation can be further divided into three sub-cases as
follows:
1. Inputs which do not fulfill conditions in the generic
plan.
2. Inputs which do not fulfill some conditions but fulfill
constraints in the generic plan.
3. Inputs which do not fulfill constraints in the generic
plan.
In the first case, consider the example system interface in
Fig. 15. Assume that the post-condition in the generic plan is
Percentage of renal damage = 95%, and this does not match
with the users input which is Percentage of renal damage =
60%. Therefore, the personalization process is terminated and
the entire planning process is deemed to have completed. As a
result, the planning system generates the output as shown in
Fig. 16 as the personalized plan for this case.

Figure 15. System interface for inputs which do not fulfill conditions in the
generic plan.

Figure 16. Personalized plan for inputs which do not fulfill conditions in the
generic plan.
In the second case, consider the example system interface
in Fig. 17. Let us assume that the post-condition in the generic
plan, i.e. Patient condition = stable (see Fig. 18), does not
match the users input which is Patient condition = not stable
(Fig. 17). Therefore, the process terminates. As a result, the
planning system generates the personalized plan as showed in
Fig. 19.

Figure 17. System interface for inputs which do not fulfill some conditions
but fulfill constraints in the generic plan.
In the third case, consider the example system interface in
Fig. 20. Let us assume that the user does not fulfill the
constraint of High blood pressure = yes, the planning system
would search the plan repository for a plan fragment which
fulfils the users input. This plan fragment is then retrieved and
Vol. 2, No. 10, 2011

66 | P a g e
used to replace the one in the generic plan which did not fulfill
the users input. Fig. 21 shows the personalized plan for this
case.

Figure 18. Generic plan for inputs which do not fulfill some conditions but
fulfill constraints in the generic plan.

Figure 19. Personalized plan for inputs which do not fufill some conditions
but fulfill constraints in the generic plan.

Figure 20. System interface for inputs which do not fulfill constraints in the
generic plan.

Figure 21. Personalized plan for inputs which do not fulfill constraints in the
generic plan.

Vol. 2, No. 10, 2011

67 | P a g e
V. DISCUSSION AND COMPARISON
In general, the generic and personalized plan generation
processes performs up to expectation. However, as a limitation
of the DP Planner system, these processes would not function
fully when a replacement cannot be found in the plan
repository. When the system encounters this situation, it will
advise the user to refer the case to a medical practitioner.
TABLE I. COMPARISON OF DP PLANNER WITH OTHER PLANNING SYSTEMS.

Table 1 shows the comparison between the DP Planner
with other planning systems, i.e. Asgaard, O-Plan, Prodigy,
STRIPS, PLANET and RAX-PS. From our observation,
Asgaard and O-Plan are established planning systems that the
DP Planner can be compared to in view that they have the
relevant plan ontology, plan representation, as well as the
planning engine for their planning system. Further comparisons
with Asgaard and O-Plan are discussed in the following sub-
sections.
A. Plan Ontology Definition and Representation
Asgaard was inspired by Belief-Desire-Intention (BDI)
model [14] while DP Planner was developed based on Goals-
Operators-Method-Selection (GOMS) model. Using the BDI
framework, Asgaard has been used to build large-scale, highly
capable agent system [15]. Therefore, Asgaard is more suited
for domains with large and complex but partly vague and
incomplete knowledge. In contrast, DP Planner is based on the
GOMS framework that has not been used to develop large-
scale systems. GOMS has been mainly used to represent
human knowledge necessary for performing certain tasks and
complex human activities. As a result, DP Planner which is
based on GOMS is more suited in domains with obvious
knowledge, i.e. knowledge that is confirmed and complete, for
performing certain tasks and knowledge.
Due to its simplicity, the DP Planner plan ontology was
developed without the need for any ontology tool such as
Protg. The DP Planner plan representation in XML is also
intuitively easy to understand. Asbru (which is Asgaards plan
representation language) uses a machine-readable language
(Backus-Naur Form or BNF syntax) to annotate guidelines
based on the task-specific ontology.
Asbru also requires the use of an ontology editor such as
Protg for the acquisition of clinical guidelines based on the
same ontology and GMT (Guideline Markup Tool) to translate
the guideline into a formal representation written in XML [16].
DP Planners ontology is also easier and simpler compared
to O-Plan which has its own detailed ontology structure. The
O-Plan plan representation is in Task Formalism form and will
change in different O-Plan agents in which it is situated. This is
quite complex compared to DP Planner in which the plan
representation remains in XML form in any situation.
B. Planning Algorithm Definition
DP Planner generates a generic plan by retrieving an
existing plan with similar characteristics to the current planning
requirements, and adapting the generic plan by reusing existing
plans to produce a personalized plan. This is akin to a case-
based approach with adaptation. Asgaard employs a similar
approach to DP Planner whereby it also applies plan adaptation
in its planning process.
However, the difference is that Asgaard adopts the
transformational type of adaptation whereas DP Planner adopts
a derivational analogy type followed by the transformational
type. Derivational analogy potentially reduces the search space
by ignoring the unnecessary choices. This is achieved using the
DP Planners similarity measurement technique. This is only
suitable in situations when most of the previous plans require
extensive adaptation and when the cost of saving rationale is
low [15]. The cost of saving rationale here means the ability to
fulfill the requirements of a particular plan fragment that was
defined as a constraint. However, it presupposes that the
derivational traces exist. In contrast, when this is not possible,
transformational analogy is the better choice because the plans
themselves can be used for adaptation. Thus, in DP Planner,
derivational analogy is applied in generic plan generation while
transformational analogy is applied in the plan personalization
in view that the cost of saving rationale is higher.
In general, Asgaard appears more robust in view that it is a
fully deployed system, and that it has a monitoring component
which monitors changes to the users situation, while DP
Planner is not a deployed system and therefore relies on users
to report any changes.
Vol. 2, No. 10, 2011

68 | P a g e
O-Plan on the other hand is based on software agents and
provides a hierarchical planning architecture to support
planning and control with temporal and resource constraint
handling [17]. O-Plan is also designed as a fully deployed
system. O-Plans architecture shows that it has five major
components which are Domain Information, Knowledge
Sources, Support Tools, Plan State, and Controller. O-Plan also
has the agent architecture since it has a Task Assignment,
Planner and Execution agent.
In O-Plan, its plan repair algorithm involves two tables (see
Fig. 22): TOME (Table of Multiple Effect), and GOST (Goal
Structure Table) [18]. An execution failure occurs when one or
more of the expected effects at a node-end fail to be asserted.
Each effect is recorded in the TOME and when an action
depends on an effect asserted earlier, it is recorded in the
GOST (Step 1).
When an execution failure occurs, the TOME will be
updated and its relation with GOST entries will be found. If it
is related with any of the GOST entries, then the Knowledge
Sources is used to fix the problem (Step 2). The Knowledge
Sources are responsible for determining the consequences of
unexpected events or of actions that do not execute as intended,
for deciding what action to take when a problem is detected,
and for making repairs to the effected plan (Step 3 and 4) [17].

Figure 22. Solving execution failure in O-Plan [12].
When comparing the DP Planners approach with that of O-
Plan, it seems that the DP Planner approach is simpler as only
two stages are needed to solve the failure whereas O-Plan
requires four stages to solve the failure (see Fig. 23).
VI. FUTURE WORK
Presently, the DP Planner is implemented in a local
environment. Its capabilities can be extended further by
deploying it in a grid environment with distributed plan
repositories and planning engines.

Figure 23. Solving execution failure in DP Planner.
For this purpose, the open source Globus Tool Kit can be
utilized to allow sharing of computing power, databases and
other tools online across corporate, institutional, and
geographic boundaries autonomously and safely. The Globus
Tool Kit is used in tandem with the Open Grid Services
Architecture with Data Access and Integration (OGSADAI)
which provides a service group registry that can be used to
identify database services that offer specific data tables [19].
In order to communicate with OGSADAI, the DP Planner
would potentially require middleware software to communicate
with the DBMS which will store the planning data. Fig. 24
shows the OGSADAI with Globus in a possible DP Planner
planning scenario. Here, the planning engines and plan
repositories are distributed across different locations and each
of these planning engines accesses the XML database
containing the plan repository (or medical treatment plans) via
Xindice (a suggested XML DBMS) while the planning results
are delivered through the Client Tool Kit middleware whereby
it provides the communication channel between the requesting
node and the processing planning engine [20].
VII. CONCLUSION
There are many types of planning systems currently
available in the literature though most are static in nature. In
this paper, we presented (1) a plan ontology and representation,
and (2) a dynamic planning engine which makes use of generic
plans and plan fragments, based on our plan ontology, as a
planning template. The processing of the planning engine is
based on similarity measure, matching, linking, and constraint
techniques. In the comparative study carried out, it was
observed that the DP Planner possesses features that rival that
of other planning systems, in particular that of Asgaard and O-
Plan. It is hoped that the DP Planner will make planning
initiative more efficient and effective in delivering applicable
plans to users especially to healthcare providers and patients.

Vol. 2, No. 10, 2011

69 | P a g e

Figure 24. The DP Planner in a grid computing environment.
ACKNOWLEDGMENT
The authors wish to thank the Ministry of Higher
Education, Malaysia and Universiti Sains Malaysia for funding
this work through the Fundamental Research Grant Scheme
(FRGS) under the project entitled A Dynamic Planning
Algorithm Using Ontologies and Constraints on the Grid.
REFERENCES
[1] N. Mahiddin, An Ontology and Constraint-based Approach for
Dynamic Personalised Planning, Master Thesis, Universiti Sains
Malaysia, 2009.
[2] A. Tate, Towards a plan ontology, Journal of the Italian Association
for AI, AI*IA Notizie, Special Issues on Aspects of Planning Research,
vol. 9, no. 1, pp. 19-26, March 1996.
[3] A. Tate, A plan ontology - a working document, in Proceedings of the
Workshop on Ontology Development and Use, La Jolle, CA, U.S.A.,
1994.
[4] Y. Shahar, S. Miksch, and P. Johnson, A task-specific ontology for
design and execution of time-oriented skeletal plans, in Tenth
Knowledge Acquisition for Knowledge-Based Systems Workshop,
Banff, Canada, 1996.
[5] S. Miksch, Y. Shahar, and P. Johnson, Asbru: a task specific, intention-
based and time-oriented language for representing skeletal plans, in
Proceedings of the Seventh Workshop on Knowledge Engineering
Methods and Languages (KEML-97), Milton Keynes, U.K., 1997.
[6] S. Miksch, A. Seyfang, and R. Kosara, Plan management: supporting
all steps of protocol development and deployment, in Proceedings of
the EUNITE-Workshop on Intelligent Systems in Patient Care, Vienna,
Austsria, pp. 35-42, 2001.
[7] A. K. Jnsson, P. H. Morris, N. Muscettola, and K. Rajan, Planning in
interplanetary space: Theory and practice, in Proceedings of the Fifth
International Conference on Artificial Intelligence Planning and
Scheduling (AIPS-2000), Breckenridge, CO, U.S.A., pp. 177-186, 2000.
[8] J. Frank, A. K. Jnsson, and P. H. Morris, On reformulating planning as
dynamic constraint satisfaction, in Proceedings of the 4th International
Symposium on Abstraction, Reformulation and Approximation, London,
U.K.: Springer-Verlag, 2000.
[9] M. S. Atkin, and P. R. Cohen, Physical planning and dynamics, in
Working Notes of the AAAI Fall Symposium on Distributed Continual
Planning, Orlando, FL, U.S.A., pp. 4-9, 1998.
[10] M. S. Atkin, and P. R. Cohen, Using simulation and critical points to
defines states continous search spaces, in Proceedings of Simulation
Conference, Orlando, FL, U.S.A., vol. 1, pp. 464-470, 2000.
[11] S. S. R. Abidi, Y. H. Chong, S. R. Abidi, An intelligent info-structure
for composing and pushing personalised healthcare information over the
internet, in Proceedings of the 14th IEEE Symposium on Computer
Based Medical Systems (CBMS 2001), Bethesda, MD, U.S.A., pp. 225-
230, 2001.
[12] D. Jonassen, M. Tessmer, and W. Hannum, Task Analysis Methods for
Instructional Design, Mahwah, NJ: Lawrence Erlbaum Associates, 1999.
[13] N. Mahiddin, Y.-N. Cheah, and F. Haron, A generic plan ontology for
dynamic health plans, in Proceedings of the International Conference of
Knowledge Engineering 2005 (IKE 05), Las Vegas, NV, U.S.A., 2005.
[14] S. Miksch, and A. Seyfang, Continual planning with time-oriented,
skeletal plans, in Proceedings of the 14th European Conference on
Artificial Intelligence (ECAI 2000), Amsterdam, The Netherlands: IOS
Press, pp. 512-515, 2000.
[15] H. M. Avila, and M. T. Cox, Case-based plan adaptation: An analysis
and review, IEEE Intelligent Systems, vol. 23, no. 4, pp.75-81, 2008.
[16] Y. Shahar, S. Miksch, and P. Johnson, The Asgaard project: A task
specific framework for the application and critiquing of time-oriented
clinical guidelines, Artificial Intelligence in Medicine, vol. 14, pp. 29-
51, 1998.
[17] O-Plan Team, O-plan: Architecture guide (version 2.3), 1995. URL:
http://www.aiai.ed.ac.uk/oplan/documents/ANY/oplan-architecture-
guide.pdf. Retrieved: 18 October 2011.
[18] B. Drabble, A. Tate, and J. Dalton, Repairing plans on-the-fly, in
Proceedings of the NASA Workshop on Planning and Scheduling for
Space, 1997.
[19] Globus Project Team, Globus tool kit homepage, 2007. URL:
http://www-unix.globus.org/toolkit/about.html. Retrieved: 18 October
2011.
[20] Globus Project Team, Software components for grid systems and
applications, 2007. URL: http://www.globus.org/grid_software.
Retrieved: 18 October 2011.
[21] Felix, A. A., & Taofiki, A. A. (2011). On Algebraic Spectrum of
Ontology Evaluation. IJACSA - International Journal of Advanced
Computer Science and Applications, 2(7), 159-168.
[22] Vanitha, K., Yasudha, K., & Soujanya, K. N. (2011). The Development
Process of the Semantic Web and Web Ontology. IJACSA -
2(7).
AUTHORS PROFILE
Normadiah Mahiddin received her B.Comp.Sc. (Hons) degree from Universiti
Sains Malaysia in 2003, and M.Sc. (Computer Science) from the same
university in 2009. She is currently a Ph.D. candidate at Universiti
Kebangsaan Malaysia. Her research interests include knowledge management,
intelligent systems, health informatics, and automated planning.
Yu-N Cheah received his B.Comp.Sc. (Hons) and Ph.D. degrees from
Universiti Sains Malaysia in 1998 and 2002 respectively. He is currently
lecturing at the School of Computer Sciences, Universiti Sains Malaysia. His
research interests include knowledge management, intelligent systems, health
informatics, and semantic technologies.
Fazilah Haron received her B.Sc. (in Computer Science) from the University
of Wisconsin-Madison, U.S.A. and her Ph.D. from the University of Leeds,
U.K. She is an Associate Professor at the School of Computer Sciences,
Universiti Sains Malaysia and currently on secondment at Taibah University,
Madinah, Saudi Arabia. Her research interests include modeling and
simulation of crowd, parallel and distributed processing, and grid computing.

Vol. 2, No. 10, 2011
70 | P a g e
Asynchronous Checkpointing and Optimistic
Message Logging for Mobile Ad Hoc Networks

Ruchi Tuli
Affiliation 1 : Research Scholar, Department of Computer
Science
Singhania University, Pacheri Bari (Rajasthan) INDIA
Affiliation 2: Department of Computer Sc. & Engg., P.O
Box 31387, Yanbu University College,
Kingdom of Saudi Arabia
Parveen Kumar

Professor,
Department of Computer Science,
Meerut Institute of Engineering & Technology, Meerut
INDIA

Abstract - In recent years the advancements in wireless
communication technology and mobile computing fueled a steady
increase in both number and types of applications for wireless
networks. Wireless networks can roughly be classified into
cellular networks which use dedicated infrastructure (like base
stations) and ad hoc networks without infrastructure. A Mobile
Ad Hoc Network (MANET) is a collection of mobile nodes that
can communicate with each other using Multihop wireless links
without using any fixed infrastructure and centralized controller.
Since this type of networks exhibits a dynamic topology, that is,
the nodes move very frequently, it is hard to establish some
intermittent connectivity in this scenario. Fault tolerance is one of
the key issues for MANETs. In a cluster federation, clusters are
gathered to provide huge computing power. Clustering methods
allow fast connection and also better routing and topology
management of mobile ad hoc networks (MANET). To work
efficiently on such systems, networks characteristics have to be
taken into account, for e.g. the latency between two nodes of
different clusters is much higher than the latency between two
nodes of the same cluster. In this paper, we present a message
logging protocol well-suited to provide fault tolerance for cluster
federations in mobile ad hoc networks. The proposed scheme is
based on optimistic message logging.
Keywords MANETs; clusterhead; checkpointing; pessimistic
logging; fault tolerance; Mobile Host.
I. INTRODUCTION
Wireless networks include infrastructure-based networks
and ad hoc networks. Most wireless infrastructure-based
networks are established by a one hop radio connection to a
wired network. On the other hand, mobile ad hoc networks are
decentralized networks that develop through self-organization
[1]. The original idea of MANET started out in the early 1970s.
At this time they were known as packet radio networks. Lately,
substantial progress has been made in technologies like
microelectronics, wireless signal processing, distributed
computing and VLSI (Very Large Scale Integration) circuit
design and manufacturing [2]. This has given the possibility to
put together node and network devices in order to create
wireless communications with ad hoc capability.
MANETs are formed by a group of nodes that can transmit
and receive data and also relay data among themselves.
Communication between nodes is made over wireless links. A
pair of nodes can establish a wireless link among themselves
only if they are within transmission range of each other. An
important feature of ad hoc networks is that routes between two
hosts may consist of hops through other hosts in the network
[3]. When a sender node wants to communicate with a receiver
node, it may happen that they are not within communication
range of each other. However, they might have the chance to
communicate if other hosts that lie in-between are willing to
forward packets for them. This characteristic of MANET is
known as multihopping. An example is shown in figure 1.
Node A can communicate directly (single-hop) with node B,
node C and node D. If A wants to communicate with node E,
node C must serve as an intermediate node for communication
between them. Therefore, the communication between nodes A
and E is multi-hop.

Figure 1 Multi-hop communication in a mobile ad hoc network
Today wireless bluetooth, personal area networks (PAN),
IEEE 802.11 a/b/g, wireless local area networks (WLAN) and
HIPERLAN/2, are communication standards that include ad
hoc features [4]. Figure 2 shows an example of a mobile ad hoc
network composed of commonly used wireless devices.

Figure 2 Wireless Mobile Ad hoc Network
Vol. 2, No. 10, 2011
71 | P a g e
Checkpoint and message logging protocols are designed for
saving the execution state of mobile application, so that when a
MH recovers from a failure, the mobile application can roll
back to the last saved consistent state, and restart execution
with recovery guarantees. The existing protocols assume that
the MHs disk storage is not stable and thus checkpoint and log
information are stored at the base stations [5], [6].
Log-based rollback recovery exploits the fact that a process
execution can be modeled as a sequence of deterministic state
intervals, each starting with the execution of a non-
deterministic event. A non-deterministic event can be the
receipt of a message from another process or an event internal
to the process. A message send event is not a non-deterministic
event. Log-based rollback recovery assumes that all non-
deterministic events can be identified and their corresponding
determinants can be logged into the stable storage.
During failure-free operation, each process logs the
determinants of all non-deterministic events that it observes
onto the stable storage. Additionally, each process also takes
checkpoints to reduce the extent of rollback during recovery.
After a failure occurs, the failed processes recover by using the
checkpoints and logged determinants to replay the
corresponding non-deterministic events precisely as they
occurred during the pre-failure execution. Because execution
within each deterministic interval depends only on the
sequence of non-deterministic events that preceded the
intervals beginning, the pre-failure execution of a failed
process can be reconstructed during recovery up to the first
non-deterministic event whose determinant is not logged. The
deterministic intervals composing the process execution are
called state intervals. The state intervals are partially ordered
by the Lamports happen-before relation [7].
Message logging techniques are classified into pessimistic
[8], optimistic [9], [10], [11] and causal [12], [13], [14], [15].
Pessimistic logging protocols assume that a failure can occur
after any non-deterministic event in the computation. This
assumption is pessimistic since in reality failures are rare. In
their most straightforward form, pessimistic protocols log to
the stable storage the determinant of each non-deterministic
event before the event affects the computation. A pessimistic
protocol is one in which each process p never sends a message
until it knows that all messages delivered before sending the
previously sent messages are logged. Pessimistic protocols will
never create any inconsistent process (orphans), and so the
reconstruction of the state of a crashed process is very
straightforward. The pessimistic protocols potentially block a
process for each message it receives.
In optimistic logging protocols, processes log determinants
asynchronously to the stable storage. These protocols
optimistically assume that logging will be complete before a
failure occurs. Determinants are kept in a volatile log, and are
periodically flushed to the stable storage. Thus, optimistic
logging does not require the application to block waiting for the
determinants to be written to the stable storage, and therefore
incurs much less overhead during failure-free execution.
However, the price paid is more complicated recovery, garbage
collection, and slower output commit. If a process fails, the
determinants in its volatile log are lost, and the state intervals
that were started by the non-deterministic events corresponding
to these determinants cannot be recovered. Furthermore, if the
failed process sent a message during any of the state intervals
that cannot be recovered, the receiver of the message becomes
an orphan process and must roll back to undo the effects of
receiving the message. Optimistic logging protocols do not
implement the always-no-orphans condition.
Causal logging combines the advantages of both pessimistic
and optimistic logging at the expense of a more complex
recovery protocol. Like optimistic logging, it does not require
synchronous access to the stable storage except during output
commit. Like pessimistic logging, it allows each process to
commit output independently and never creates orphans, thus
isolating processes from the effects of failures at other
processes. Moreover, causal logging limits the rollback of any
failed process to the most recent checkpoint on the stable
storage, thus minimizing the storage overhead and the amount
of lost work. Causal logging protocols make sure that the
always-no-orphans property holds by ensuring that the
determinant of each non-deterministic event that causally
precedes the state of a process is either stable or it is available
locally to that process.
In this paper we focus on optimistic based message logging
for communications in a clustered ad hoc network, since the
checkpointing-only schemes are not suitable for the mobile
environment and also in ad hoc environments in which
unreliable mobile hosts and fragile network connection may
hinder any kind of coordination for checkpointing and
recovery. In order to cope with the storage problem, the task of
logging is assigned to the CH instead of MHs, since each
message heading to a MH is routed through the CH. Also, in
order to reduce the overhead imposed on mobile hosts, cluster
heads take charge of logging and dependency tracking, and
mobile hosts maintain only a small amount of information for
mobility tracking.
The rest of this paper is organized as follows: Section 2
discusses related work and problem formulation. In section 3,
system model is described. Section 4 explains the basic
algorithm and comparison with existing schemes is described
in section 5. Finally section 6 concludes the paper.

II. RELATED WORK AND PROBLEM FORMULATION
A. Related Work

Application failure recovery in the mobile computing
environment has received considerable attention in the recent
years. The schemes that have been proposed employ
checkpointing, logging or a combination of both, recognizing
the inherent limitations of the mobile computing environments.
Since the requirements of a ad hoc network are different from
the mobile computing environment, these issues needs to be
addressed in this area as well.
Prakash and Singhal describe in [16] a checkpointing
algorithm for Mobile Conputing System. Checkpoint collection
is synchronous and non-blocking. A minimum number of
nodes are forced to take checkpoints. Each MH maintains a
dependence vector. MHs maintain causal relationships through
message. This scheme reduces energy consumption by
Vol. 2, No. 10, 2011
72 | P a g e
powering down individual components during periods of low
activity.
In [17] T.Park et.al has presented an efficient movement
based recovery scheme. This scheme is a combination of
message logging and independent checkpointing. Main feature
of this algorithm is that a host carrying its information to the
nearby MSS can recover instantly in case of a failure. To
enhance failure-free execution, concept of a 'certain range' is
introduced. An MH moving inside a range , recovery
information remains in host MSS otherwise it moves recovery
information to nearby MSS. Though recovery is ensured,
failure-free execution cost increases. Due to this out of range
concept overheads due to transfer of checkpoint from one MSS
to another MSS increases many fold.
Sapna E. George [18]et.al describes a checkpointing and
logging scheme based on mobility of MHs. A checkpoint is
saved when hand-off count exceeds a predefined optimum
threshold. Optimum threshold is decided as a function of MH's
mobility rate, failure rate and log arrival rate. Recovery
probability is calculated and recovery cost is minimized in this
scheme.
Acharya et al. [6] describes uncoordinated checkpointing,
where multiple MHs can arrive at a global consistent
checkpoint without coordination messages. However, neither it
takes into account how failure recovery is achieved nor does it
address the issue of recovery information management in the
face of MH movement.
In [19] authors proposed an independent checkpointing
scheme which saves the state of processes in the computer to
which a mobile host is currently attached.
The authors in [20] presents a low overhead recovery
scheme based on a communication induced checkpointing,
which allows the processes to take checkpoints asynchronously
and uses communication-induced checkpoint coordination for
the progression of the recovery line. The scheme also uses
selective pessimistic message logging at the receiver to recover
the lost messages. However, the recovery scheme can handle
only a single failure at a time.
P. Kumar and A. Khunteta [22] proposed a minimum-
process coordinated checkpointing algorithm for deterministic
mobile distributed systems, where no useless checkpoints are
taken, no blocking of processes takes place, and anti-messages
of very few messages are logged during checkpointing. In their
algorithm they have tried to reduce the loss of checkpointing
effort when any process fails to take its checkpoint in
coordination with others.
B. Problem Formulation
Cluster federations are hierarchical systems. The latency
between two clusters is much higher than the latency between
two nodes of the same cluster. For efficient execution on such
systems, applications must take into account the topology of
the cluster federation. Communications between nodes of the
same cluster should be favored over communications between
nodes of different clusters.
The objective of the present work is to design an optimistic
based message logging for communications in a clustered ad
hoc network, since the checkpointing-only schemes are not
suitable for the mobile environment and also in ad hoc
environments. In order to cope with the storage problem, the
task of logging is assigned to the CH instead of MHs, since
each message heading to a MH is routed through the CH. In
order to reduce the size of dependency information carried in
each message for asynchronous recovery, only the messages
between the CHs carry the information, and the dependency
between the MHs residing in the same Cluster can be traced
through the message order within the CH. Using the restricted
dependency tracking, no extra overhead is imposed on MHs.
Of course, there is a possibility of unnecessary rollbacks due to
the imprecise dependency information, however, comparing
with the checkpointing-only schemes, the chance of rollback
propagation in the message logging schemes is very low.
III. SYSTEM MODEL
A successful approach for dealing with the maintenance of
mobile ad hoc networks is by partitioning the network into
clusters. In this way the network becomes more manageable.
Clustering is a method which aggregates nodes into groups.
These groups are contained by the network and they are known
as clusters. Clusters are analogous to cells in a cellular network.
However, the cluster organization of an ad hoc network cannot
be achieved offline as in fixed networks [21]. In most
clustering techniques nodes are selected to play different roles
according to a certain criteria. In general, three types of nodes
are defined:
Ordinary nodes :- Ordinary nodes are members of a
cluster which do not have neighbors belonging to a different
cluster.
Gateway nodes:- Gateway nodes are nodes in a non-
clusterhead state located at the periphery of a cluster. These
types of nodes are called gateways because they are able to
listen to transmissions from another node which is in a different
cluster. To accomplish this, a gateway node must have at least
one neighbor that is a member of another cluster.

Clusterheads:- Most clustering approaches for mobile
ad hoc networks select a subset of nodes in order to form a
network backbone that supports control functions. A set of the
selected nodes are called clusterheads and each node in the
network is associated with one. Clusterheads are connected
with one another directly or through gateway nodes. The union
of gateway nodes and clusterheads form a connected backbone.
This connected backbone helps simplify functions such as
channel access, bandwidth allocation, routing power control
and virtual-circuit support.

Clusterheads are analogous to the base station concept in
current cellular systems. They act as local coordinators in
resolving channel scheduling and performing power control.
However, the difference of a clusterhead from a conventional
base station resides in the fact that a clusterhead does not have
special hardware, it is selected among the set of stations and it
presents a dynamic and mobile behavior. Since clusterheads
must perform extra work with respect to ordinary nodes they
can easily become a single point of failure within a cluster. For
this reason, the clusterhead election process should consider for
the clusterhead role, those nodes with a higher degree of
Vol. 2, No. 10, 2011
73 | P a g e
relative stability. The main task of a clusterhead is to calculate
the routes for long-distance messages and to forward inter-
cluster packets. Figure 3 shows the system model and different
roles of nodes in a mobile ad hoc network organized by
clusters.

Figure: - 3 System Model
The clustering system considered in this paper follows the
model presented in the figure above The system is organized
into various clusters, each having a clusterhead and ordinary
nodes, which will be termed as Mobile Hosts; a set of dynamic
links can be established between a MH and a Cluster head. The
area covered by a cluster head is called a cell. A MH residing
in a cluster can be connected to the clusterhead servicing the
cluster and the MH can communicate to another MH only
through the cluster head. The links in the dynamic network
support FIFO communication in both directions.
For a MH to leave a cluster and enter into another cluster, it
first has to end its current connection by sending a leave(x)
message to the cluster head, where x is the sequence number of
the last message received from the cluster head, and then
establish a new connection by sending join (MH-id, previous
cluster-id, previous clusterhead-id) message to the new cluster
head in the new cluster. Each cluster head maintains a list of
identifiers, called a Current_Nodelist, with which nodes it
connected at current time
A MH can also disconnect itself from the cluster head
voluntarily without leaving the cluster by sending a disconnect
(x) message to conserve power. When the cluster head receives
a disconnect message from a Node, it marks that node to be
disconnected by setting a flag that maintains a list of
voluntarily disconnected MHs, called disconnected_Nodelist
Later on, the MH can reconnect to any cluster by sending a
reconnect (MH-id, previous cluster-id, previous
clusterhead_id) message to the current cluster head. If the MH
is reconnected to a new cluster, the new cluster head informs
the previous cluster head of the reconnection so that the
previous cluster head can perform the proper hand-off
procedures.
IV. PROPOSED ALGORITHM
The ordinary Nodes in the clustered ad hoc networks are
considered highly vulnerable to failures, while the cluster heads
are relatively reliable as they are chosen among the nodes with
a higher degree of relative stability. By reliable CH, we mean
that recovery information for MHs can never be lost due to its
own failure. With this assumption, the volatile memory space
of a CH can be utilized as a stable storage to save checkpoints
and message logs of MHs.
A. Data Structures and Notations
Chk
i
x
= checkpoint sequence number
i=0..n
j=0..n
MH
i
= No. of Mobile hosts
i=0..n
CH = Cluster head
m
i
rcv
=No. of messages a mobile host has received
i=0..n
rcv=0..n
Locate
i
=variable to retrieve information after failure of MH
chk_seq=sequence number of latest checkpoint
cp_loc=current ID of a cluster
cp_ch=current cluster head ID
msg_seq =sequence number of the first message logged after
checkpoint
log_set=IDs of cluster heads that store MH logs
timeToCkp
i
=Timer to take checkpoint on MH
B. Checkpointing and Message Logging

Each mobile host MH independently takes a checkpoint and
a unique sequence number is assigned to each checkpoint. For
the checkpointing, MH first saves its current state as a
checkpoint and then transfers the checkpoint to CH to which it
is currently connected. Chk
i
x
denotes the xth checkpoint of
MH
i
. Each checkpoint is identified by a pair of (i, x). MH
i
then
sends the checkpoint to its current CH, say CH
p
. Each Mobile
host also maintains a variable m
i
rcv
, to count the number of
messages a mobile host has received, and the value of m
i
rcv
is
sent with the checkpoint to the cluster head. On the receipt of a
checkpoint and other related information CH saves it into
stable storage. The value of m
i
rcv
sent with the checkpoint to
CH is used to decide the correct position of the checkpoint with
respect to logged messages.
Each CH
p
also maintains the message log for the nodes
residing in that cluster. Since each message that is delivered to
MH in the cluster is routed through CH, logging of messages
incurs little overhead. Let Msg
i
x
denotes xth message delivered
to MH
i
. In addition, CH
p
also logs the messages related to the
mobility of the nodes, such as join, leave, disconnect and
reconnect. Any of these messages sent from MH must also
carry the value of m
i
rcv
and sequence number is logged with
message.
A node after a failure should be able to locate its latest
checkpoint and the message log for recovery. Each mobile host
also contains a variable Locate
i
, to retrieve the information
after failure. Locate
i
contains chk_seq, cp_loc, cp_ch, msg_seq
and log_set. chk_seq stores the sequence number of latest
checkpoint and cp_loc stores the ID of the cluster and cp_ch
stores the ID of the CH that has recorded the latest checkpoint.
Let this cluster be called cluster
in
and CH be called CHp.
msg_seq denotes the sequence number for the first message
logged after the checkpoint and log_set contains the IDs of the
cluster heads that stores its logs. At every checkpoint, cp_loc is
updated with the current CH, cp_seq is updated with the
Vol. 2, No. 10, 2011
74 | P a g e
sequence number of the latest checkpoint and log_set is
cleared. At every logging activity, the IDs of the current CHs
are added to log_set if it is not present already. Locate
i
is
logged by the CHs which a MH has visited. When a mobile
host joins or reconnects to a new cluster, say CH
p
, it sends
Locate
i
with the connection message. Also when mobile host
disconnects itself from or leaves CH
p
, it sends Locate
i
with
disconnection message if information in the Locate
i
has been
changed since the connection was established. CH
p
on receipt
of Locate
i
logs it with the message. Each mobile host also
maintains a variable timeToCkp
i
which defines Time interval
until next checkpoint
C. Algorithm
We describe pseudocode for the checkpointing and message
logging protocol here

D. Checkpointing at MH

If ( timeToCkp
i
= Expire) then
chk_seq
i
=chk_seq
i
+1; //increment checkpoint
sequence number
Perform checkpointing , Chk
i

chk_seqi

Save (i, chk_seq
i
, m
i
rcv
,) with Chk
i

chk_seqi
// updating the Locate field

Locate
i
.chk_seq=chk_seq
i
;
Locate
i
.cp_loc= cluster
in
;
Locate
i
.cp_ch=p;
Locate
i
.msg_seq= m
i
rcv
+1;
Locate.log_set=NULL;
Send (Chk
i

chk_seqi
[i, chk_seq, m
i
rcv
]) to CH
p

Else
Continue computation;
If (CH
p
=recv Chk
i

chk_seqi
) from MH
Save (Chk
i

chk_seqi
[i, chk_seq, m
i
rcv
,]);
Else
Continue computation
E. Message logging at CH

a) When cluster head delivers a message M, to Mobile
host
msg_seq
i
=msg_seq
i
+1;
Insert (M
i
msg_seqi
[i, msg_seq
i
]) into Log;

b) When Mobile host receives a message from cluster
head (CHp)
If(p Locate
i
.log_set)
Locate
i
.log_set=Locate
i
.log_set p;
c) When Mobile host sends a message to Cluster head
If (M {join, leave, disconnect, reconnect})
Send (M [Locate
i
, m
i
r_seq
])
d) When Cluster head receives a message M, from
mobile host
If (M {join, leave, disconnect, reconnect})
Insert (M (Locate
i
, m
i
rcv
)) into log;
F. Proof of Correctness

Theorem I :- If a MH fails, its state can be reconstructed
independently
Proof :- Let MH
i
state be [
si
0
, s
i
1
, s
i
2
.s
i
l
) before failure,
which indicates messages e
i
o
, e
i
x-1
, e
i
x
, .e
i
y
, where 1y, e
i
x
is
the first message from the last checkpoint and e
i
y
is the last
message before failure. Since all the messages delivered to
MH
i
are logged in CH and Locate
i
.log_set indicates the order in
which MH
i
has contacted CH since its last checkpoint. After a
failure MH
i
should rollback to the latest checkpoint and the
logged messages in the same order and it can reconstruct the
same state intervals as the ones before failure. Because all the
messages sent and received events are recorded, the MHis
state can be reconstructed.
V. HANDLING FAILURES AND DISCONNECTIONS
We distinguish here failures and disconnections. Failures
can be categorized as mobile host falls and is damaged, lost
or stolen, battery is discharged. Disconnections are termed as
hand-off. Since the mobility rate of mobile hosts in ad hoc
networks is very high, so while connected a mobile host can
change its position and can join another cluster. This movement
is termed as disconnection. We will discuss these two issues
separately.
After a MH recovers from a failure, either a mobile host is
in the same cluster or the cluster can be changed. When a MH
i

recovers from a failure, it first sends a join(MH-id, previous
cluster-id, previous clusterhead-id) to its current CH, say CH
p
.
CH
p
checks it Active_nodelist list and Disconnected_nodelist.
If MH
i
is found in any of these lists that mean MH after a
failure has recovered in current cell and thus the Loacte
i
must
have been logged in CH
p
.

Sometimes it may happen that CH is not able to find MHi
either in Active_nodelist list or Disconnected_nodelist, which
means that MH after a failure has moved to another cluster. In
this case, firstly MH sends a join (MH-id, previous cluster-id,
previous clusterhead-id) to its current CH, say CH
q
. Now, CH
q

will broadcast the recovery message, so that previous CH
which has been contacted by MH
i
can deliver the most recent
Loacte
i
to CH
q
. After CH
q
receives the most recent Locate
i
, it
starts with the recovery procedure. During the recovery of MH
i
,
new messages heading to MH
i
can be logged at CH
q
. However,
these messages are delivered to MH
i
after consuming all the
messages in the log. Only the MHs which have failed rollback
to the latest checkpoint and replay the logged messages to
ensure the global recovery line and no other MHs need to
rollback.
A node after a disconnection should be able to locate its
latest checkpoint and the message log for recovery. Each
mobile host contains a variable Locate
i
, to retrieve the
information after failure which we have discussed in section
4.1. For each hand-off or disconnection, a MH within a cluster
transfers the checkpoint and message logs to the current cluster
head, so that the recovery information can be retrieved later
from the cluster head. For a MH, MH
i
, connected to the cluster
head CH
p
, in cluster CL
i
first saves its checkpoints and
message logs and updates the information in Locate
i
. Let
Vol. 2, No. 10, 2011
75 | P a g e
MH_data
(i, p)
denotes the checkpoints and message logs of MH
i

saved by CH
p
of CL
i
. When MH
i
leaves CL
i
and joins another
cluster head say CH
new
of cluster CL
k
, a hand-off procedure is
initiated by CH
new
sending a handoff-request for MH
i
to CH
p
.
While the hand-off procedure is performed, the recovery
information is transferred from CH
p
to CH
new
. Figure 4 depicts
the sequence of events that take place for the recovery
information transfer.

Figure 4 : Handling failures and disconnections
VI. PERFORMANCE COMPARISON
In [1] authors proposed a communication pattern based
checkpointing scheme to save consistent global states, in which
a checkpoint is taken whenever a message reception is
preceded by a message transmission. An independent
checkpointing scheme which saves the state of processes in the
computer to which a mobile host is currently attached was
proposed by authors in [12]. Neither of the above approaches
needs the checkpointing coordination, however, they may
enforce a large number of checkpoints. [13] proposed a low-
cost synchronous checkpointing scheme, in which a process
can advance its checkpoint asynchronously, however, it may
result in considerable message overhead and an inconsistency.
[10] presents a low overhead recovery scheme based on a
communication induced checkpointing, which allows the
processes to take checkpoints asynchronously and uses
communication-induced checkpoint coordination for the
progression of the recovery line. The scheme also uses
selective pessimistic message logging at the receiver to recover
the lost messages. However, the recovery scheme can handle
only a single failure at a time. In [18] authors describe a
checkpointing and logging scheme based on mobility of MHs.
A checkpoint is saved when hand-off count exceeds a
predefined optimum threshold. Optimum threshold is decided
as a function of MH's mobility rate, failure rate and log arrival
rate. Recovery probability is calculated and recovery cost is
minimized in this scheme.
We have described an optimistic based message logging
scheme since checkpointing-only schemes are not suitable for
ad hoc environments and most of the schemes described above
are based on checkpointing-only approach. Also, we have
followed the asynchronous checkpointing approach, as
asynchronous recovery is desirable in ad hoc environments in
which MH can be disconnected any time from the network and
co-ordination may not be possible.

VII. CONCLUSION
In this paper, we have proposed an optimistic based
message logging approach for cluster based ad hoc networks in
which each MH in the cluster takes checkpoint independently.
Also, each message that is delivered to MH in the cluster is
routed through CH which avoids the overhead of message
logging at MH. MH only carries minimum information and all
the dependency tracking and mobility of MH can be properly
traced by CH. The asynchronous checkpointing scheme relives
the MH from any kind of coordination and they can take their
checkpoints whenever they want.
REFERENCES
[1] C. Prehofer, C. Bettstetter. Self organization in communication
networks: Principles and design paradigms. IEEE Communications
Magazine. Vol. 43. Issue 7. 2005. pp. 78-85.
[2] Y. P. Chen, A. L. Liestman, J. Liu. Ad Hoc and Sensor Networks,
Wireless Networks and Mobile Computing. Clustering Algorithms for
Ad hoc Wireless Networks. Vol. 2. Chapter 7: Nova Science Publishers,
Hauppage NY, 2004. pp. 145-164. 7. J. Wu, J. Cao. Connected k-hop
clustering in ad hoc networks. ICPP. 2005. pp 373-380.
[3] I. Chatzigiannakis, S. Nikoletseas. Design and analysis of an efficient
communication strategy for hierarchical and highly changing ad-hoc
mobile networks. Mobile Networks and Applications. Vol. 9. 2004. pp.
248-263.
[4] M. Frodigh, P. Johansson, P. Larsson. Wireless Ad Hoc Networking--
The Art of Networking without a Network. Ericsson Review. Vol. 77.
2000. pp. 248-263.
[5] S. Gadiraju, Vijay Kumar, Recovery in the mobile wireless
environments using mobile agents, IEEE Transactions on Mobile
Computing, June 2004, Vol. 3.
[6] A. Acharya , B. R. Badrinath, Checkpointing distributed applications
on mobile computers, in Proc. 3rd Int. Conf. Parallel and Distributed
Information Systems, Austin, Texas, 1994, pp. 7380.
[7] L. Lamport. Time, Clocks, and the Ordering of Events in a Distributed
System. Communications of the ACM, 21(7):558565, 1978.
[8] D. B. Johnson and W. Zwaenpoel, Sender-Based Message Logging, In
Digest of Papers:17th International Symposium on Fault-Tolerant
Computing, pp. 14-19, 1987.
[9] O. P. Damani and V. K. Garg, How to Recover Efficiently and
Asynchronously when Optimism Fails, In Proc. the 16th International
Conference on Distributed Computing Systems, pp. 108-115, 1996.
[10] D. B. Johnson and W. Zwaenepoel, Recovery in distributed systems
using optimistic message logging and checkpointing, In Proc. the 7th
Annual ACM Symposium on Principles of Distributed Computing, pp.
171-181, 1988.
[11] R. B. Strom and S. Yemeni, Optimistic recovery in distributed systems,
ACM Transactions on Computer Systems, Vol.3, No.3, pp. 204-226,
1985.
[12] L. Alvisi, B. Hoppe, and K. Marzullo, Nonblocking and Orphan-Free
Message Logging Protocols, In Proc. the 23th Symposium on Fault-
Tolerant Computing, pp. 145-154, 1993.
[13] L. Alvisi and K. Marzullo, Message Logging: Pessimistic, Optimistic,
Causal and Optimal,IEEE Transactions on Software Engineering,
Vol.24, No.2, pp. 149-159, 1998.
[14] E.Elnozahy, On the relevance of Communication Costs of Rollback
Recovery Protocols, In Proc. the 15th ACM Symposium on Principles
of Distributed Computing, pp. 74-79, 1995.
[15] E. N. Elnozahy and W. Zwaenepoel, Manet: Transparent rollback-
recovery with low overhead, limited rollback and fast output commit,
IEEE Transactions on Computers, Vol.41, No.5, pp. 526-531, 1992.
[16] R.Prakash, M.Singhal, (1996) Low Cost Checkpointing and Failure
Recovery in Mobile Computing Systems, IEEE Transacrions on
Parallel and Distrinuted Systems, VOL. 7, NO. IO, OCTOBER 1996
Vol. 2, No. 10, 2011
76 | P a g e
[17] Taesoon Park, Namyoon Woo, Heon Y. Yeom,(2003) An Efficient
recovery scheme for fault tolerant mobile computing systems, Future
Generation Computer System, 19(1): 37-53
[18] Sapna E. George,Ing-Ray Chen,Ying Jin, (2006)Movement-Based
Checkpointing and Logging for Recovery in Mobile Computing
Systems, MobiDE, 51-58
[19] D.K. Pradhan, P. Krishna, and N.H. Vaiday. Recoverable mobile
environment : Design and trade-off analysis. In Proc. of the 26th Intl
Symp. on Fault Tolerant Computing, 1996.
[20] D. Manivannan and M. Singhal. Failure recovery based on quasi-
synchronous checkpointing in mobile computing systems. OSU-CISRC-
796-TR36, Dept. of Computer and Information Science, The Ohio State
University, 1996.
[21] A. B. McDonald, T. F. Znati. A mobility-based framework for adaptive
clustering in wireless Ad Hoc networks. IEEE Journal on Selected
Areas in Communications. Vol. 17. 1999. pp. 1466-1487.
[22] Parveen Kumar, Ajay Khunteta, Anti-message Logging based
coordinated checkpointing protocol for Deterministic Mobile Computing
Systems, International Journal of Computer Applications (0975-887),
Vol. 3-No. 1, June, 2010.
[23] Parveen Kumar, A Low-Cost Hybrid Coordinated Checkpointing
Protocol for Mobile Distributed Systems, Mobile Information Systems
[An International Journal from IOS Press, Netherlands] pp 13-32, Vol. 4,
No. 1, 2007. [Listed in ACM Portal & Science Citation Index Expanded]
[24] Lalit Kumar, Parveen Kumar A Synchronous Checkpointing Protocol
for Mobile Distributed Systems: A Probabilistic Approach,
International Journal of Information and Computer Security [], pp 298-
314, Vol. 3 No. 1, 2007. [An International Journal from Inderscience
Publishers, USA, Listed in ACM Portal]
[25] Sunil Kumar, R K Chauhan, Parveen Kumar, A Minimum-process
Coordinated Checkpointing Protocol for Mobile Computing Systems,
International Journal of Foundations of Computer science,Vol 19, No. 4,
pp 1015-1038 (2008). [ Listed in Science Citation Index Expanded ]
[26] Parveen Kumar, Lalit Kumar, R K Chauhan, A Hybrid Coordinated
Checkpointing Protocol for Mobile Computing Systems, IETE journal
of research, Vol. 52, Nos 2&3, pp 247-254, 2006. [Listed in Science
Citation Index Expanded ]
[27] Lalit Kumar, Parveen Kumar, R.K. Chauhan, Logging based
Coordinated Checkpointing in Mobile Distributed Computing Systems,
IETE Journal of Research, vol. 51, no. 6, pp. 485-490, 2005. [Listed in
Science Citation Index Expanded ]
[28] Obaida, M. A., Faisal, S. A., & Roy, T. K. (2011). AODV Robust
( AODV R ): An Analytic Approach to Shield Ad-hoc Networks from
Black Holes. IJACSA - International Journal of Advanced Computer
Science and Applications, 2(8), 97-102.
[29] Journal, I., Science, A. C., & Hod, M. (2011). A Survey on Attacks and
Defense Metrics of Routing Mechanism in Mobile Ad hoc Networks.
IJACSA - International Journal of Advanced Computer Science and
Applications, 2(3), 7-12.
[30] Indukuri, R. K. R. (2011). Dominating Sets and Spanning Tree based
Clustering Algorithms for Mobile Ad hoc Networks. IJACSA -
2(2).
Vol. 2, No. 10, 2011

77 | P a g e
An Experimental Improvement Analysis of Loss
Tolerant TCP (LT-TCP) For Wireless Network

Md. Abdullah Al Mamun
Dept. of CSE, DUET
Gazipur 1700, Bangladesh
Momotaz Begum
Lecturer, Dept. of CSE, DUET, Gazipur 1700 Bangladesh
Sumaya kazary
Asst. Prof., Dept. of CSE, DUET, Gazipur 1700 Bangladesh
Md. Rubel
Dept. of CSE, DUET
Gazipur 1700, Bangladesh
AbstractNow-a-days TCP is a famous protocol used in Internet
but the main problem is packet losses due to congestion. In this
thesis we proposed a new Loss Tolerant TCP (LT-TCP), an
enhancement of TCP which makes it robust and applicable for
extreme wireless environment. In the proposed LT-TCP two
additional term, data and data header compression are added in
existing LT-TCP. We reduce the total volume of data and packet
size in our adaptive method which is able to minimize the
congestion and increase the reliability of wireless communication.
The ECN respond about random data packet loss and disruption
process. The overhead of Forward Error Control FEC is imposed
just-in-time process and target to maximize the performance
even if the path characteristics are uncertain. This proposal show
that it will perform better over regular TCP and it is possible to
reduce packet losses up to 40-50%.
Keywords- LT-TCP; Mix reliability; Timeouts; Congestion
avoidance; end-to-end Algorithms; Compression technique; Packet
erasure rate; TCP SACK; RFEC; PFEC; RAR; ECN; throughput;
RTT; Goodput.
I. INTRODUCTION
TCP is the most popular protocol to deliver data reliably
regardless of the form and construction of the network [1].
When a data packet traverse a wireless link, a major fraction of
packet losses due to transmission error [2]. TCP is robust in
that it can adapt to disparate network conditions [4, 7]. LT-TCP
uses an adaptive, end-to-end hybrid Automatic Repeat
request/Forward Error Correction ARQ/ FEC [3, 4] reliability
strategy and exploits Explicit Congestion Notification ECN for
incipient congestion detection [5]. Congestion control is the
major problem of managing network traffic where the total
demand of resources such as bandwidth among the computing
users exceeds the available capacity [8]. In terms of packet loss
is occurred more often due to high Bit Error Rates (BERs) than
due to congestion [9]. When using TCP over wireless network,
it considers each packet loss as a sign of congestion and
invokes congestion control measures at the source [6]. This
results in severe performance degradation. To improve the TCP
performance such as WLAN using congestion response, mix of
reliability mechanism, time out avoidance and handle a large
volume of data [10]. It studies as extensive literature on the
performance of TCP and emphasis the ways distinguish the
effects due to congestion. It address the TCP congestion
avoidance and control issue over the wireless links from the
end-to-end communication, thereby leading to efficient
network resource utilization and improving the application
response time. It also analyze widely used TCP end-to-end
algorithm and presents a new technique that enables the TCP to
better adapt to the wireless environment [11].
We provision a data compression technique which reduce
the volume of data [12], hence reduce the total number of
packets and proactive FEC in the original window as a function
of the estimate of the actual packet erasure rate. Subsequently
reactive FEC is used to mitigate the effects of erasures, during
the transmission phase. An adaptive maximum segment size
(MSS) provides a minimum number of packets in the TCP
window, again seeking to risk of timeouts [10]. We seek to
adaptively balance the FEC and packet overhead. While
reducing the risk timeouts and also rapidly erased packets [3].
II. PROPOSED LT-TCP
A. Overview of Flow Graph
In accordance to Fig: 1 application data are providing
compression technique which yield compressed data. This
compressed data broken into TCP segments where the MSS is
chosen to accommodate proactive FEC (PFEC) packets in the
window. Reactive FEC (RFEC) packets are computed at the
same time and held in reserve. Feedback from the receiver
provides not only the loss estimate but also information (e.g.
SACK blocks) that can be used to compute the number of
RFEC packets to send for each block [10]. When the sender
receives ACK, it determines the type of packets to send
(Data/PFEC/RFEC) and transmits them. This provides self-
clocking and follows the semantics of TCP behavior. LT-TCP
comprises the following building blocks that complement each
other and extend SACK to provide resilience.PFEC and RFEC
help to data recovery. PFEC operates in conjunction with
adaptive MSS and determined by the current estimates of loss
rate. RFEC is computed based on feedback from the receiver
and the loss rate estimate. The receiver can reconstruct the data
packets as soon as any k out of n packets arrives at the receiver.
B. Data and Header Compression
Compression is a technique which reduces the volume of
original information [7, 12]. This is done either to reduce the
volume of size of text and images to be transmitted or to reduce
the bandwidth that is required for its transmission audio and
Vol. 2, No. 10, 2011

78 | P a g e
video. It required less bandwidth during transmission it is also
secured because of its encrypted form.
C. RAR Compression Technique
RAR is a powerful allowing you to manage and control
archive files.
Data
Type
Original
Data
Size
Compressed
Data Size
Number Of Segment
(MSS 1500 Bytes)
Before After
Text 10 KB 4.3 KB 7 4
10 MB 3.5 MB 6990 2567
100 MB 47.9 MB 69905 33745
Image 100 KB 69.5 KB 65 46
1MB 789 KB 687 535
10 MB 8 MB 6989 5800
Audio 10 MB 8.9 MB 6991 5786
100 MB 89.6 MB 69905 62633
1 GB 985 MB 715800 688565
Video 10 MB 8.9 MB 6990 6222
100 MB 91.9 MB 69905 64733
1 GB 999 MB 715698 697444

Figure 1. Compression comparison in terms of data size
Compressed output for different data type [12] Console
RAR [12] supports archives only in RAR format, the names of
which usually have a .rar extension. ZIP and other format are
not supported.

Figure 2. Proposed LT-TCP Flow graph

0
20
40
60
80
100
120
Text Image Audio Video
D
a
t
a

S
i
z
e

(
M
B
)
Data type
Original Data
Size
Compressed
Data Size
Vol. 2, No. 10, 2011

79 | P a g e
Figure 3. Compression comparison in terms of number of packets.
D. Van Jacobson TCP/IP Header Compression
The IPv4 header is 20 bytes and when carrying UDP (8
bytes) and RTP (12 bytes) at least the packet header become 40
bytes. A compression scheme usually compresses such header
to 2-4 bytes. On and average, considering a few uncompressed
packets and few relatively large packets, more than 80%
savings [7] can be observed. When the compared with the
payload being carried, in such cases as voice type where
payload size is usually static in range of 20-60 bytes, the header
size presents a huge overhead. In header compression in such
cases results is major bandwidth savings. The IPv6 with a
header size of 40 bytes is gaining wide acceptance and has
been included in Release 5 and onward version of 3G wireless
network. In this case, header compressions yield in even more
savings.

Figure 4. Abstract view of header compression [13].
TABLE I. THE HEADER COMPRESSION GAINS [13].
Protocol
Headers
Total
header size
(bytes)
Min. compressed
header size (bytes)
Compression
gain (%)
IPV4/TCP 40 4 90
IPV4/UDP 28 1 96.4
IPv4/UDP/RTP 40 1 97.5
IPv6/TCP 60 4 93.3
IPv6/UDP 48 3 93.75
IPv6/UDP/RTP 60 3 95
These benefits lead to improved QoS in the network and the
possibility for operators to improve their average revenue per
user. The operators will be able to retain and attract customers
with better QoS on the network and more services and content
the links.
Finally, complete content and organizational editing before
formatting. Please take note of the following items when
proofreading spelling and grammar:
III. PERFORMANCE ANALYSIS
A. Data & Header Compression Comparison
We use single bottleneck test case fig 5 with 10 flows and
erasure rates varying from 0-50%. Host is ECN enabled;
bottleneck implements RED/ECN on a 250 KB buffer (i.e. up
to 500 packets of size of 500 bytes). minthres and maxthres
values are as shown. According to table the 1, 2 and 3 we
observed the performance for uncompressed data and
compressed data & header both are given below
TABLE II. PERFORMANCE OF UNCOMPRESSED DATA
Parameter LT-TCP
RTT 10 ms 80 ms 200 ms
Goodputs (MB/s) 5.4 4.5 3.9
Number of timeout 17 10 5
Throughput (MB/s) 7.4 6.3 5.1
TABLE III. PERFORMANCE OF COMPRESSED DATA
Parameter LT-TCP
Goodputs (MB/s) 7.3 6.5 5.1
Number of timeout 12 7 3
Throughput (MB/s) 9.5 7.3 6.4

From the above tables we observed that we compressed the
data and header it mitigate the congestion and packet losses.
The flow congestion is lasting for 100s for each operation. To
assess the contribution of LT-TCP components, we use a 30%
PER test case. Metrics include aggregate throughput, goodputs,
number of timeouts and congestion window dynamics. We
account for all packet header overheads.
B. Proposed LT-TCP vs. Traditional LT-TCP
In terms of packet erasure both perform well. However, the
performance of traditional LT-TCP drops to 10% sometimes
even more. In addition an error rate (40%) is sufficient to break
a single LT-TCP connection due to repeated timeouts but
proposed LT-TCP can reduce it significantly. Fig 6 shows
different error rates for a number of RTT that means the
degradation of performance is linear. Proposed LT-TCP
manages packet error rates by avoiding timeouts and able to
recover lost packets using proactive and reactive FEC even if
RTT is too high.
C. Performance Evaluation
Finally we observe the Goodputs of both LT-TCP are
shown in table IV. As we can see the proposed LT-TCPs
Goodputs is higher than the traditional LT-TCP and
performance is increasing with respect in time. It perform
maximum at 80ms.
TABLE IV. PERFORMANCE OF PROPOSED LT-TCP
Parameter LT-TCP
LT-TCP 5.4 4.5 3.9
Proposed LT-TCP 7.3 6.5 5.1
Increased performance (%) 19 20 12

0
10000
20000
30000
40000
50000
60000
70000
80000
Text Image Audio Vedio
N
u
m
b
e
r

o
f

p
a
c
k
e
t
s
Data type
Number of
original packets
Number of
Compressed
packets
Vol. 2, No. 10, 2011

80 | P a g e

0
1
2
3
4
5
6
7
8
10 ms 80 ms 200 ms
G
o
o
d
p
u
t
s
(
M
b
p
s
)
Round Trip Time
LT-TCP
Proposed LT-
TCP

Figure 5. LT-TCP performance with increase erasure rate and RTT

Figure 6. Performance graph of proposed LT-TCP
IV. CONCLUSION
Our TCP is the dominant reliable protocol used in internet;
we have proposed a Loss-Tolerant TCP (LT-TCP) which
introduces additional mechanisms as data & header
compression in an adaptive manner. Our enhancement allow
good performance even under demanding conditions through
reduce the total volume of data that made less number of
packet and packet size which mitigate congestion and perform
better than traditional LT-TCP.
REFERENCES
[1] Computer Networks by Andrew S. Tanenbum.
[2] TCP PERFORMANCE ENHANCEMENT OVER WIRELESS
NETWORKS By Aiyathurai
ir.canterbury.ac.nz/bitstream/10092/1229/1/thesis_fulltext.pdf.
[3] S. Biswas, G.Judd, D. Aguayo, J. Bicket and Morris: Link-level
measurements from an 802.11b mesh network. In SIGCOMM, Aug02
[4] C. Ladas, R. M. Edwards AMIEE, M. Mahdavi, and G. A. Manson
(2002): Class based selective ARQ Scheme for high performance TCP
and UDP over wireless links. Paper presented at the mobile and wireless
network, 2002. 4th international workshop on it.
[5] K. K. Ramkrishnan S. Floyed and D. Black: The addition of Explicite
Congestion Notification (ECN) to IP, Sept 2001.
[6] C. Barakat and E. Altman: Bandwidth tradeoff between TCP and link
level FEC. Computer Networks 39(2): 133-150, June 2002.
[7] S. Casner and V. Jacobson: Compressing IP/UDP/RTP Headers for low
speed serial links. RFC 2508, Feb 1999. IETF Network working group.
[8] RFC 793 (1981). Transmission control protocol. From
http:/www.ietf.org/rfc/rfc793.txt.
[9] Data Communication and Networking by behrouz A. Forouzan.
[10] Omesh Tickoo, Vijaynarayanan Subramanian, Shivkumar Kalyanaraman
and K. K. Ramakrishnan: LT-TCP: End-to-End Framework to Improve
TCP Performance over Networks with Lossy Channels. In thirteen
international workshop on Quality of Service (IWQoS 2005), Passau,
Germany 2005.
[11] CHIU D. M and JAIN R.: analysis of increase and decrease algorithms
for congestion avoidance in computer networks. Computer Networks
and ISDN system, Vol 17 June, pp. 1-14
[12] Compression principles techniques and tools RAR.
[13] White paper: An introduction to IP header compression from EFFNET
AB.
AUTHORS PROFILE
Md. Abdullah al mamun obtained his Bachelor of Science in
Engineering degree from Department of Computer Science
and Engineering (CSE) of Dhaka University of Engineering
& Technology (DUET), Gazipur-1700, Bangladesh. At
present he is performing extensive research on Network and
Web Security, Wireless networking, Software Architecture,
Machine Vision, Artificial Intelligence, Protocol
cryptography. His key research interest includes Cryptography, analysis and
Algorithm designs, E-mail: [email protected].

Momotaz Begum, Lecturer, Department of CSE, DUET and
she achieved her B. Sc. in Engineering degree from
Department of Computer Science and Engineering (CSE),
DUET, Bangladesh. E-mail: [email protected]

Sumaya kazary, Asst. Prof., Department of CSE, DUET and
she achieved her B. Sc. in Engineering degree from
Department of Computer Science and Engineering (CSE),
DUET, Bangladesh. E-mail: [email protected]

0
1
2
3
4
5
6
7
8
9
10
0 10 20 30 40 50
G
o
o
d
p
u
t
s

(
M
b
p
s
)

packet error rate (%)
RTT: 6ms
RTT: 24ms
RTT: 104ms
Vol. 2, No. 10, 2011

81 | P a g e
Comparison of Workflow Scheduling Algorithms in
Cloud Computing

Navjot Kaur
CSE Department, PTU Jalandhar
LLRIET
Moga, India
Taranjit Singh Aulakh
CSE Department , PTU Jalandhar
BGIET
Sangrur, India
Rajbir Singh Cheema
IT Department, PTU Jalandhar
LLRIET
Moga, India

Abstract Cloud computing has gained popularity in recent
times. Cloud computing is internet based computing, whereby
shared resources, software and information are provided to
computers and other devices on demand, like a public utility.
Cloud computing is technology that uses the internet and central
remote servers to maintain data and applications. This
technology allows consumers and businesses to use application
without installation and access their personal files at any
computer with internet access. The main aim of my work is to
study various problems, issues and types of scheduling algorithms
for cloud workflows as well as on designing new workflow
algorithms for cloud Workflow management system. The
proposed algorithms are implemented on real time cloud which is
developed using Microsoft .Net Technologies. The algorithms are
compared with each other on the basis of parameters like Total
execution time, Execution time for algorithm, Estimated
execution time. Experimental results generated via simulation
shown that Algorithm 2 is much better than Algorithm 1, as it
reduced makespan time.
Keywords- Cloud Computing; Workflows; Scheduling; Makespan;
Task ordering; Resource Allocation.
I. INTRODUCTION
Cloud computing is Internet-based computing, whereby
shared resources, software and information are provided to
computers and other devices on-demand, like a public utility.
Cloud computing is a technology that uses the internet and
central remote servers to maintain data and applications. Cloud
computing allows consumers and businesses to use applications
without installation and access their personal files at any
computer with internet access. This technology allows for
much more efficient computing by centralizing storage,
memory, processing and bandwidth.
A. Workflows
The WfMC (Workflow Management Coalition) defined a
workflow as the automation of a business process, in whole or
part, during which documents, information or tasks are passed
from one participant to another for action, according to a set of
procedural rules.
WfMC published its reference model in [1], identifying the
interfaces within this structure which enable products to
interoperate at a variety of levels. This model defines a
workflow management system and the most important system
interfaces (see Fig 1).

Fig. 1 WfMCs Workflow Reference Model
1) Workflow Engine. A software service that provides the
run-time environment in order to create, manage and execute
workflow instances.
2) Process Definition. The representation of a workflow
process in a form which supports automated manipulation.
3) Workflow Interoperability. Interfaces to support
interoperability between different workflow systems.
4) Invoked Applications. Interfaces to support interaction
with a variety of IT applications.
5) Workflow Client Applications. Interfaces to support
interaction with the user interface.
6) Administration and Monitoring. Interfaces to provide
system monitoring and metric functions to facilitate the
management of composite workflow application environments.
It can be seen that scheduling is a function module of the
Workflow Engine(s), thus it is a significant part of workflow
management systems.
The rest of the paper is structured as follows: Related work
is discussed in Section II. Then section III describes our
Proposed Work. The Implementation is presented in Section
IV. And Section V will show the experimental details and
simulation results. Finally Section VI includes the future scope
of our research work.
II. RELATED WORK
A. Cloud Platforms
A comprehensive survey of cloud computing is defined
by number of researchers. There are no. of definitions of cloud
Process
Definition
Interface 1
Workflow API and Interchange

Workflow
Engine(s)
Workflow Enactment Service

Administratio
n &
Monitoring
Tools

I
n
t
e
r
f
a
c
e

5

Workflow
Engine(s)
Other Workflow
Enactment Service

I
n
t
e
r
f
a
c
e

4

Interface 2 Interface 3
Workflow
Client
Application
Invoked
Applications
Vol. 2, No. 10, 2011

82 | P a g e
computing. According to R. Buyya and S.Venugopal[5] Cloud
computing is defined as a type of parallel and distributed
system consisting of a collection of inter-connected and
virtualized computers that are dynamically provisioned and
presented as one or more unified computing resources based on
service-level agreements established through negotiation
between the service provider and consumers.
Sun Microsystems [3] takes an inclusive view that there
are many different types of clouds like public cloud, private
cloud, hybrid cloud .Many different applications that can be
built by using these different clouds.
Recently, several academic and industrial organizations
have started investigating and developing technologies and
infrastructure for Cloud Computing.
B. Workflow Management Systems
Workflow is concerned with the automation of procedures
whereby files and data are passed between Participants
according to a defined set of rules to achieve an overall goal. A
workflow management system defines, manages and executes
workflows on computing resources. Workflow Scheduling:
workflow scheduling is a kind of global task scheduling as it
focuses on mapping and managing the execution of inter-
dependent tasks on shared resources that are not directly under
its control. Workflow management includes five dimensions:
time, cost, fidelity, reliability and security.
The related work done in workflow management system is
shown below in tabular form (see Table III):
TABLE III
SURVEY ON WORKFLOW MANAGEMENT SYSTEM
S.
No.
Citation Brief Introduction About Paper
1. Authors: S.
Elnikety, E.
Nahum, J.
Tracey, W.
Zwaenepoel
Year: 2004
This paper [10] consider workflows that
are invoked via web requests. The
workflows are part of a web application
that spans multiple resources in the grid.
2. Authors: Jia Yu,
Rajkumar Buyya
& Chen Khong
Tham
Year: 2005
In this paper, a cost-based workflow
scheduling algorithm is proposed [11] that
minimizes execution cost while meeting
the deadline for delivering results.
3. Author: E.
Deelman, G.
Singh, D.S. Katz
Year: 2005
Pegasus [12], is proposed which is a
framework that maps complex scientific
workflows onto distributed resources such
as the Grid. DAGMan, together with
Pegasus, schedules tasks to Condor
system.
4. Author: Jia Yu,
Rajkumar Buyya
Year: 2006
A budget constraint based scheduling is
proposed [13], which minimizes execution
time while meeting a specified budget for
delivering results. A new type of genetic
algorithm is developed to solve the
scheduling optimization problem and the
scheduling algorithm is tested in a
simulated Grid tested.
5. Author: Patel,
Y. Darlaington
Year: 2006
According to this paper, there are two
categories of workflow scheduling [14].
The first category is based on the real time
data such as waiting time in the queue or
the shortest remaining execution length.
The second category is based on average
metrics such as mean arrival time, or mean
execution length.
6. Author: P.
Patala, K. G.
Shin, X. Zhu, M.
Uysal, Z. Wang,
S. Singal
Year: 2007
A control system is developed that adjusts
the resource sharing among applications to
ensure the desired QoS and maintains the
high resource utilization [15].
7. Author: J. Yu &
R. Buyya
Year: 2007
The work presented in this paper defines
two major types of workflow scheduling
[16], best-effort based and QoS constraint
based scheduling, primarily for grid
workflow management systems.
8. Author:
Zhifeng Yu &
Weisong Shi
Year: 2008
In this paper, a planner-guided strategy is
presented for multiple workflows [17]. It
ranks already tasks and decides which task
should be scheduled.
9. Author: Ke Liu,
Jin Jun chin, Yun
Yang & Hai Jin
Year: 2008
A throughput maximization strategy is
proposed [18] for scheduling transaction
intensive workflows. But it is designed for
transaction intensive workflows not for
multiple workflows.
10. Author: Meng
Xu, Lizhen Cui,
Haiyang Wang,
Yanbing Bi
Year: 2009
Authors of this paper worked on Multiple
workflow and Multiple QOS.A strategy
[19] is implemented for Multiple
Workflow Management system with
multiple QOS. The Scheduling access rate
is increased by using this strategy.
11. Author: Boris
Mejias, Peter
Van roy
Year: 2010
In this paper, researchers proposed an
architecture to organize a set of mini-
clouds provided by different institutions,
in order to provide a larger cloud that
appears to its users as a single one[20].
12. Author: J.
Kosinska, J.
Kosinski, K.
Ziehnski
Year: 2010
Purpose of this paper is to discuss various
forms of mapping cluster topology
requirements into cloud environments to
achieve higher reliability & scalability of
application executed within cloud
resources[21].
13. Author: M.
Jensen,
J.schwenk, J.M.
Bohli, L.L.
Iacono
Year: 2011
This paper initiates discussion by
contributing a concept which achieves
secutrity merits by making use of multiple
distinct clouds at the same time[22].
III. PROPOSED WORK
This section presents a set of scheduling algorithms, based
on Time management [23]. The aim of the algorithms is to
optimize the makespan, which is defined as the maximum time
taken for the completion of all the tasks in a given application.
The proposed algorithms are implemented using a service
based cloud and comparative results are shown.
The problem of scheduling a set of tasks to a set of
processors can be divided into two categories:
Job scheduling
Job mapping and scheduling
In the former category, independent jobs are to be
scheduled among the processors of a distributed computing
system to optimize overall system performance. In contrast, the
mapping and scheduling problems requires the allocation of
multiple interacting tasks of a single parallel program in order
to minimize the completion time on the parallel computer
system.
Vol. 2, No. 10, 2011

83 | P a g e
To generate the schedule, our technique is based on the
traditional list scheduling approach in which we construct a list
and schedule the nodes on the list one by one to the processors.
A. Algorithm 1
The design of our algorithm 1 is basis on the following
heuristics. It is based on the POSEC method [23]. POSEC is an
acronym for Prioritize by Organizing, Streamlining,
Economizing and Contributing. The objective of our
algorithms is efficient time management and load balancing.
There are Four Quadarnts of Descion Making :
a) Level 1: Low Urgency & Low Importance
b) Level 2: Low Urgency & High Importance
c) Level 3: High Urgency & Low Importance
d) Level 4: High Urgency & High Importance
There are Four Quadarnts of Descion Making : It needs
two types of Priority Scores to take descion , Urgency Score
and Importance Score. Urgency Score given by Cluster
Member of cloud. Importance Score is given by Cloud
Resources Manager .
Urgency Score is Calculated on the scale of 10 on the basis
of the following table.
TABLE IIV
CLASSIFICATION OF URGENCY LEVELS
Level
of
Severity
Description of Severity Initial
Respo
nse
Withi
n
Score
Level I Production application
down or major
malfunction causing
business revenue loss
resulting in majority of
users unable to perform
their normal functions
1 hour 7.6<=Score<=10
Level II Critical loss of
application functionality
or performance resulting
in high number of users
unable to perform their
normal functions
4
hours
5.1<=Score<=7.5
Level III Moderate loss of
application functionality
or performance resulting
in multiple users
impacted in their normal
functions
8
hours
2.6<=Score<=5.0
Level IV Minor loss of application
functionality or product
feature question
24
hours
1>=Score<=2.5
Importance Score is Calculated by the Resource manager
and its also on the scale of 10. The various parameter of
resource cheking are CPU time. Threades etc. we have use
resource monitor program to generate the importance score.
High Importance means the Resources are available. Low
Importance means based the Resources are Not available .
It is assumed that job consist of tasks. The cloud scheduler
assigns these tasks to resources. Also it is assumed that each
computational resource can run one application at a time, and
must run that application to completion.
Let T be a set of n tasks and m is the number of
computational resources in a cloud. We define a schedule of T
as follows: A schedule S of T onto a cloud with m resources is
a finite set of tuples<v, p, t> where v is the schedule, t is the
starting time, and p is the resource.
To generate the schedule, our technique is based on the
traditional list scheduling approach in which we construct a list
and schedule the nodes on the list one by one to the processor.
The list is constructed by ordering the jobs according to their
urgency score s. The list is static therefore the order of nodes
on the list will not change during the resource allocation
process.
We restrict ourselves to non-preemptive schedules where a
job once started has to run to completion on the same machine.
Scheduler has information about all resources such as
processing speed (in MIPS), processing cost per second, baud
rate(communication rate) and resource load during peak hours
and off peak hours.
After gathering the details of user jobs, the system
calculated the importance score. The jobs are executed on the
values of urgency and importance score.
The time management parameters used by the algorithms
are:
1) Total Execution Time: The total time consumed by the
algorithm to execute all the jobs.
2) Execution Time of Algorithm: This is the time taken by
the algorithm to execute.
3) Estimated Execution Time: Based on the average of
total execution time parameters of previous jobs.
The proposed algorithm comprises of two parts as
explained below.
A. Task Ordering Procedure, to get the schedule list
B. Resource Allocation Procedure, which allocates
resources to the jobs contained in scheduling list, generated by
task ordering procedure.
a) Task Ordering Procedure
Begin
Step 1: The list is initialized to be an empty list. The cloud
clients calculate the urgency score according to the severity of
jobs.
Step 2: The urgency score is calculated. The urgency score
is based on the scale of 10.
There are 4 cases to determine the urgency score of the job.
If Level 1: 7.6 <=Score<=10
If Level II: 5.1 <=Score<=7.5
If Level III: 2.6 <=Score<=5.0
Vol. 2, No. 10, 2011

84 | P a g e
If Level IV: 1>=Score<=2.5
Step 3. According to the urgency score , the alert templates
have set up
b) Resource Allocation Procedure:
Begin
Step 1: The Cloud Scheduler collects resources and their
characteristics like processing speed, processing cost per
second, resource load during peak hours and off peak hours.
Step 2: It generates importance score according to these
characteristic on the scale of 10.
Step 3: After collecting the information about the job
parameters like urgency and score, the jobs are executed
according to the following case.
If Urgency High, Importance High: The email alert sent
immediately.
If Urgency High, Importance low: whenever the
resources are free, the email is sent on high priority basis.
If Urgency Low, Importance High: The email alert is sent
after emptying the job Queue.
If Urgency low, Importance low: whenever the resources
are free, and the job queue is empty, the email is sent with
lower priority basis.
B. Algorithm 2
The second algorithm is based upon the Pareto Analysis
[23].
Pareto Analysis
This is the idea that 80% of tasks can be completed in 20%
of the disposable time. The remaining 20% of tasks will take up
80% of the time. This principle is used to sort tasks into two
parts. According to this form of Pareto analysis it is
recommended that tasks that fall into the first category be
assigned a higher priority.
The 80-20-rule can also be applied to increase productivity:
it is assumed that 80% of the productivity can be achieved by
doing 20% of the tasks. If productivity is the aim of time
management, then these tasks should be prioritized higher.
For example, look at your to do list- if you have 10 tasks on
there then two of those tasks will yield 80% of your results.
Alternatively, 80% of income is owned by 20% of people - it
works both ways!
The Pareto principle holds across business, academia,
politics, and a number of other areas. The foundation of this
time management skill is that: 20% of tasks yield 80% of
results
This algorithm is also comprised of two parts.
1. Task Ordering Procedure, to get the schedule list
2. Resource Allocation Procedure, which allocates
resources to the jobs contained in scheduling list, generated by
task ordering procedure.
a) Task Ordering Procedure
Begin
Step 1: The list is initialized to be an empty list. The cloud
clients send the jobs to the cloud manager according to their
priority.
Step 2: The urgency score is calculated. The urgency score
is based on the scale of 10.
There are 4 cases to determine the urgency score of the job.
If Level 1: 7.6 <=Score<=10
If Level II: 5.1 <=Score<=7.5
If Level III: 2.6 <=Score<=5.0
If Level IV: 1>=Score<=2.5
Step 3. According to the urgency score , the alert templates
have set up
b) Resource Allocation Procedure:
Step 1: From the previous set of jobs, importance score is
calculated. The score is based upon the following method:
Let t = (time to execute jobs/estimated time) %
If t = 100, then the cloud resources are utilized properly. A
high score of importance is send by the algorithm.
It t <100 and t>80, the cloud manages resources are
overloaded; but according to 80:20 rule by paleto, a high
importance score is generated.
If t<80, a low importance score is generated.
If t> 100, the cloud manager resources are underutilization,
a high importance score is generated by the cloud.
Step 2: According to the importance score and urgency
score, calculated, the jobs are executed by the cloud manager.
If Urgency High, Importance High: The email alert sent
immediately.
If Urgency High, Importance low: whenever the
resources are free, the email is sent on high priority basis.
If Urgency Low, Importance High: The email alert is sent
set after emptying the job Queue.
If Urgency low, Importance low: whenever the resources
are free, and the job queue is empty, the email is sent with
lower priority basis.
IV. IMPLEMENTATION
The management and scheduling of resources in Cloud
environment is complex, and therefore demands sophisticated
tools for analysis the algorithm before applying them to the real
system. But there are no good tools are available that serve our
needs. So we develop a service based cloud using Microsoft
.Net Technologies. The proposed algorithms are implemented
upon this real time cloud.
Feature of This cloud:
Vol. 2, No. 10, 2011

85 | P a g e
a) To send real time email alerts to the cloud clients
members like Bank, Insurance, and Hospital etc.
b) The algorithms are tested on real time cloud.
c) Googles SMTP server is used to send the mails.
d) The Database is saved on the Web server.
e) The cloud is working online. You need no special
software to test the results.
f) Visual studio 2008 is used as frontend and SQL 2005
is used as Backend.
Cloud Architecture:

Fig. 2 Architecture of service based cloud
The cloud architecture is based upon the real time email
alert system. It sends email alerts to its cluster client members.
Feature of the cloud architecture are:
a) This scenario based cloud has real life application of
sending email alerts to Bank clients, Hospital clients, and
Insurance company clouds. This cloud takes jobs from the all
other clients with their urgency score. The cloud manager
executes the jobs according to the importance score based on
cloud resources.
b) The data Flow between all the clouds is using XML.
c) XML is hardware and software free technology
d) Its widely suited for cloud application.
e) There are 4 domains used in this service based cloud.
f) Moreover its three tier architecture, the database is
stored on the other server and web services execute on the
other server.
g) FileZilla client FT\P application is used to
upload/download data from the server.
V. EXPERIMENTAL RESULTS
This section describes the experiment results obtained after
implementing the scheduling algorithms. The algorithms are
implemented in Microsoft .Net framework using a service
based cloud. It takes as input the required set of resources and a
set of tasks. The algorithms are compares with each other on
set of parameters like Total execution Time, Execution time for
Algorithm, Estimated Execution Time.
By Graphical Analysis of Experimental results, we
analysed the simulation results using graphs. Graphical data
consist of 3 cloud clients data that are scheduled by main
cloud manager. Each algorithm is run by 8 times to conclude
the results:
Experiment Results of Algorithm1:
TABLE V
EXPERIMENTAL RESULTS OF ALGORITHM 1
Jobs Total
Execution
Time
Total
Algorithm
Time
Estimated
Time
1 22419.5038 432.2809 16624
2 17782.3016 575.8664 17204
3 18689.8762 868.9554 17248
4 48342.7388 19863.5854 17385
5 21839.9645 51.0341 19449
6 21720.2652 45.0113 19599
7 28639.296 329.7595 19723
8 22714.3025 123.876 20509

Fig. 3 Line Chart results of Algorithm 1
Experimental Results of Algorithm2:
TABLE VIII
EXPERIMENTAL RESULTS OF ALGORITHM II
Jobs Total
Execution
Time
Total
Algorithm
Time
Estimated
Time
1 15048.9774 219.5119 17120
2 14730.2754 255.9244 16861
3 17153.3982 1454.7952 17256
4 17731.0295 319.7149 17359
5 23109.6739 489.1577 20219
6 19565.4316 76.3057 20371
7 22035.7285 286.8222 20331
8 22558.6223 230.3577 20412

Total
Execution
Time
Total
AlgoTime
Estimated
Time
Vol. 2, No. 10, 2011

86 | P a g e

Fig. 4 Line Chart results of Algorithm 2
Makespan comparison of Algorithm 1 and Algorithm 2:
The following Graph shows the comparison of makespan
time between the two proposed algorithms.

Fig. 5 Pie Chart with Makespan Comparison
VI. FUTURE SCOPE
We would like to extend these algorithms to include various
parameters like options for advance reservation, preemptive
jobs as well. Also, in the Future we can add more clouds to this
main cloud, to distribute the load work. Currently this cloud
provides services like email alerts, we can also extend to store
online data and providing the synchronization mechanism in
this.
REFERENCES
[1] Workflow Management Coalition, Workflow Management Coalition
Terminology & Glossary, February 1999.
[2] Ian Foster, Yong Zhao, Ioan Raicu and Shiyong Lu, Cloud Computing
and Grid Computing 360-Degree Compared, Grid Computing
Environments Workshop 2008(GCE '08).
[3] Sun Microsystems,Inc.Introduction to Cloud Computing Architecure
Whitepaper,Ist Edition,june 2009.[Online] Availabe: www.sun.com
[4] Rajkumar Buyya, Chee Shin Yeo, Srikumar Venugopal, James Broberg,
and Ivona Brandic, Cloud Computing and Emerging IT Platforms:
Vision, Hype, and Reality for Delivering Computing as the 5th Utility,
Future Generation Computer Systems, Elsevier Science, Amsterdam,
June 2009, Volume 25, Number 6, pp. 599-616
[5] R. Buyya, C. S. Yeo, and S. Venugopal, Market-Oriented Cloud
Computing: Vision, Hype, and Reality for Delivering IT Services as
Computing Utilities, Proceedings of the 10th IEEE International
Conference on High Performance Computing and Communications
(HPCC 2008, IEEE CS Press, Los Alamitos, CA, USA), Sept. 25-27,
2008, Dalian, China.
[6] Microsoft Azure. http://www.microsoft.com/azure/ [30 Oct. 2010]
[7] Sun network.com (Sun Grid). http://www.network.com [18 July 2010]
[8] Amazon Elastic Compute Cloud (EC2). http://www.amazon.com/ec2/ [18
July 2010]
[9] Google App Engine. http://appengine.google.com [18 July 2010]
[10] S. Elnikety, E. Nahum, J. Tracey and W. Zwaenepoel, "A method for
transparent admission control and request scheduling in e-commerce
web sites," in WWW '04: Proceedings of the 13th International
Conference on World Wide Web, 2004, pp. 276-286.
[11] Jia Yu, Rajkumar Buyya and Chen Khong Tham, Cost-based
Scheduling of Scientific Workflow Applications on Utility Grids, In 1st
IEEE International Conference on e-Science and Grid Computing,
Melbourne, Australia, Dec. 5-8, 2005.
[12] E. Deelman, G. Singh, M.-H. Su, J. Blythe, Y. Gil, C. Kesselman, G.
Mehta, K. Vahi, G. B. Berriman, J. Good,A. Laity, J. C. Jacob, and D. S.
Katz. Pegasus: A framework for mapping complex scientific workflows
onto distributed systems. Sci. Program., 13(3):219237, 2005.
[13] Jia Yu and Rajkumar Buyya, A Budget Constrained Scheduling of
Workflow Applications on Utility Grids using Genetic
Algorithms,Proceedings of the 15th IEEE International symposium on
High Performance Distributed Computing (HPDC 2006), IEEE CS
Press,Los Alamitos, CA, USA, June 19-23, 2006, Paris, France.
[14] Patel, Y. Darlington, J., "A novel stochastic algorithm for scheduling
QoS-constrained workflows in a web service-oriented grid," in Web
Intelligence and International Agent Technology Workshops, 2006. WI-
IAT 2006 Workshops. 2006 IEEE/WIC/ACM International Conference
on, pp. 437-442.
[15] P. Padala, K. G. Shin, X. Zhu, M. Uysal, Z. Wang, S. Singhal, A.
Merchant and K. Salem, "Adaptive control of virtualized resources in
utility computing environments," SIGOPS Oper. Syst. Rev., vol. 41, pp.
289-302, 2007.
[16]J. Yu and R. Buyya, Workflow Schdeduling Algorithms for Grid
Computing, Technical Report, GRIDS-TR-2007-10, Grid Computing
and Distributed Systems Laboratory, The University of Melbourne,
Australia, May 2007.
[17] Zhifeng Yu and Weisong Shi, "A Planner-Guided Scheduling Strategy
for Multiple Workflow Applications," icppw, pp.1-8, International
Conference on Parallel Processing - Workshops, 2008.
[18] Ke Liu, Jinjun Chen, Yun Yang and Hai Jin, A throughput maximization
strategy for scheduling transaction-intensive workflows on SwinDeW-
G, Concurrency and Computation:Practice and Experience, Wiley,
20(15):1807-1820, Oct.2008.
[19] Meng Xu, Lizhen Cui, Haiyang Wang, Yanbing Bi, "A Multiple QoS
Constrained Scheduling Strategy of Multiple Workflows for Cloud
Computing," ispa, pp.629-634, 2009 IEEE International Symposium on
Parallel and Distributed Processing with Applications, 2009.
[20] Boris Mejias, Peter Van Roy, From Mini-clouds to Cloud
Computing,2010
[21] J. Kosinska, J. Kosinski, K, Zielinski, The Concept of Application
Clustering in Cloud Computing Environments, 2010
[22] M. Jensen, J. Schwenk, J.M. Bohli, L.L. Iacono, Security prospects
through cloud computing by adopting Multiple Clouds, 2011
[23] Wikipedia .http://en.wikipedia.org/wiki/Time_management [25 march
2011].
[24] Rao, N. M. (2010). Cloud Computing Through Mobile-Learning.
IJACSA - International Journal of Advanced Computer Science and
Applications, 1(6).
[25] Goel, M. S., Kiran, R., & Garg, D. (2011). Impact of Cloud Computing
on ERP implementations in Higher Education. IJACSA - International
Journal of Advanced Computer Science and Applications, 2(6), 146-148.
Total
Execution
Time
Total
AlgoTime
Estimated
Time
Series1,
Makespan
Time For
Algorithm
1,
25268.531
, 57%
Series1,
Makespan
Time For
Algorithm
2,
18991.642
1, 43%
Makespan Comparison
Vol. 2, No.10, 2011

87 | P a g e
Study of Indian Banks Websites for Cyber Crime
Safety Mechanism

Susheel Chandra Bhatt
Research Scholar, Computer Science department
Kumaun University, Nainital, Uttarakhand, India
Durgesh Pant
Prof. & Director, School of Computer Science & IT
Uttarakhand Open University
Dehradun Campus, Dehradun, Uttarakhand, India

AbstractThe human society has undergone tremendous changes
from time to time with rapid pace at social level from the
beginning and technological level ever since the rise of
technologies. This technology word changes the human life in
every manner and every sector. Banking field is one of them.
Banking in India originated in the last decades of the 18th
century. Since that time the banking sector applying different
ways to provide facilities and securities to a common man
regarding to money. Security issues play extremely important
role in the implementation of technologies specially in banking
sector. Further on it becomes more critical when it comes to the
cyber security which is at the core of banking sector. After the
arrival of Internet and WWW this banking sector is totally
change specially in terms of security because now money is in
your hand on a single click. Now user has number of choices to
manage his money with different kind of methods. In this paper
an attempt has been made to put forward various issues of Indian
banks websites for cyber-crime safety mechanism.
Keywords- Cyber; Encryption; Phishing; Secure Socket Layer.
I. INTRODUCTION
Crime is a social and economic phenomenon and is old as
the human society. Crime is a legal concept and has the
sanction of the law. Crime or an offence is a legal wrong that
can be followed by criminal proceedings which may result
into punishment[1]. The hallmark of criminality is that, it is
breach of the criminal law. Per Lord Atkin the criminal
quality of an act cannot be discovered by reference to any
standard but one: is the act prohibited with penal
consequences[2]. Cyber-crime is a term used to broadly
describe criminal activity in which computers or computer
networks are a tool, a target, or a place of criminal activity and
include everything from electronic cracking to denial of
service attacks. It is also used to include traditional crimes in
which computers or networks are used to enable the illegal
activity. Cyber-crime is the latest and perhaps the most
complicated problem in the cyber world. Any criminal activity
that uses a computer either as an instrumentality, target or a
means for perpetuating further crimes comes within the ambit
of cyber-crime [3]. Cyber-crimes are computer related as well
as computer generated crimes which are increasing day by
day. Cyber-crime is a term used to broadly describe criminal
activity in which computers or computer networks are a tool, a
target, or a place of criminal activity and include everything
from electronic cracking to denial of service attacks. It is also
used to include traditional crimes in which computers or
networks are used to enable the illicit activity.
Phishing is a way of attempting to acquire sensitive
information such as usernames, passwords and credit card
details by illegally as a trustworthy entity in an electronic
communication. Communications purporting to be from
popular social web sites, auction sites, online payment
processors or IT administrators are commonly used to lure the
unsuspecting public. Phishing is typically carried out by e-
mail spoofing or instant messaging [4], and it often directs
users to enter details at a fake website whose look and feel are
almost identical to the legitimate one. Phishing is an example
of social engineering techniques used to deceive users [5] and
exploits the poor usability of current web security
technologies. Attempts to deal with the growing number of
reported phishing incidents include legislation, user training,
public awareness, and technical security measures. A phishing
technique was described in detail in 1987, and the first
recorded use of the term "phishing" was made in 1996. The
term is a variant of fishing [6] probably influenced
by phreaking [7][8] and alludes to "baits" used in hopes that
the potential victim will "bite" by clicking a malicious link or
opening a malicious attachment, in which case their financial
information and passwords may then be stolen. Not all
phishing attacks require a fake website. Messages that claim to
be from a bank, tell the users to dial a phone number regarding
problems with their bank accounts [9]. Once the phone
number (owned by the phisher, and provided by a Voice over
IP service) was dialled, it prompts the user to enter the account
number and PIN. Vishing (voice phishing) sometimes uses
fake caller-ID data to give the appearance that calls is from a
trusted organization [10].
II. CYBER CRIME SAFETY MECHANISM
A. Password encryption
One of the most important security features used today are
passwords. It is important for users to have secure, strong
passwords. Most of the more recent Linux distributions
include password programs that do not allow user to set an
easily guessable password. User has to make sure
the password program is up to date and has these features.
Encryption is very useful, possibly even necessary in this day
and age. There are all sorts of methods of encrypting data,
each with its own set of characteristics. Most Unixes (and
Vol. 2, No.10, 2011

88 | P a g e
Linux is no exception) primarily use a one-way encryption
algorithm, called DES (Data Encryption Standard) to encrypt
the passwords. This encrypted password is then stored in
database. When user attempt to login, the password user type
in is encrypted again and compared with the entry in the file
that stores the passwords. If they match, it must be the same
password, it allowed access. Although DES is a two-way
encryption algorithm (user can code and then decode a
message, given the right keys), the variant that most Unixes
use is one-way. This means that it should not be possible to
reverse the encryption to get the password from the contents
of database.
B. Virtual Keyboard
A virtual keyboard is a computer keyboard that a user
operates by typing on or within a wireless- or optical-
detectable surface or area rather than by depressing physical
keys. Such a system can enable the user of a small handheld
device, such as a cellular telephone or a PDA (personal digital
assistant) to have full keyboard capability. In one technology,
the keyboard is projected optically on a flat surface and, as the
user touches the image of a key, the optical device detects the
stroke and sends it to the computer. In another technology, the
keyboard is projected on an area and selected keys are
transmitted as wireless signals using the short-
range Bluetooth technology. Theoretically, with either
approach, the keyboard could even be projected in space and
the user could type by moving fingers through the air. The
term virtual keyboard is sometimes used to mean a soft
keyboard , which appears on a display screen as an image
map . In some cases, a software-based keyboard can be
customized. Depending on the host system and specific
software, the user (who may be someone unable to use a
regular keyboard) can use a touch screen or a mouse to select
the keys. Virtual keyboard can be categorized as:
virtual keyboards with touch screen keyboard layouts
or sensing areas [11]
optically projected keyboard layouts or similar
arrangements of "keys" or sensing areas[12][13]
optically detected human hand and finger motions[14]
C. Secured Socket Layer
Secure Sockets Layer (SSL), are cryptographic protocols
that provides communication security over the Internet [15].
SSL is the standard security technology for establishing an
encrypted link between a web server and a browser. This link
ensures that all data passed between the web server and
browsers remain private and integral. SSL is an industry
standard and is used by millions of websites in the protection
of their online transactions with their customers. To be able to
create an SSL connection a web server requires an SSL
Certificate. When users choose to activate SSL on web server
he will be prompted to complete a number of questions about
the identity of the website and the company. The web server
then creates two cryptographic keys - a Private Key and a
Public Key. The Public Key does not need to be secret and is
placed into a Certificate Signing Request (CSR) - a data file
also containing the details. User should then submit the CSR.
During the SSL Certificate application process, the
Certification Authority will validate the details and issue an
SSL Certificate containing the details and allowing user to use
SSL. The web server will match issued SSL Certificate to
Private Key. The web server will then be able to establish an
encrypted link between the website and the customer's web
browser. The complexities of the SSL protocol remain
invisible to the customers. Instead their browsers provide them
with a key indicator to let them know they are currently
protected by an SSL encrypted session - the lock icon in the
lower right-hand corner, clicking on the lock icon displays the
SSL Certificate and the details about it. All SSL Certificates
are issued to either companies or legally accountable
individuals. Typically an SSL Certificate will contain domain
name, company name, address, city, state and country. It will
also contain the expiration date of the Certificate and details of
the Certification Authority responsible for the issuance of the
Certificate. When a browser connects to a secure site it will
retrieve the site's SSL Certificate and check that it has not
expired, it has been issued by a Certification Authority the
browser trusts, and that it is being used by the website for
which it has been issued. If it fails on any one of these checks
the browser will display a warning to the end user letting them
know that the site is not secured by SSL.
D. SMS Alerts
SMS as used on modern handsets was originated
from radio telegraphy in radio memo pagers using
standardized phone protocols and later defined as part of
the Global System for Mobile Communications (GSM) series
of standards in 1985 [16] as a means of sending messages of
up to 160 characters[17] to and from GSM mobile
handsets[18]. SMS stands for short message service.
An SMS alert is a message sent to a cellular device, such as a
phone, to notify the receiver of something. An SMS alert is
received in much the same way as a phone call is received.
There is normally a sound or vibration that will indicate that
the message has come in. There are various types
of SMS alerts that people may consent to. These include
appointment reminders, banking transactions, and specials or
sales offered by businesses they patronize. In many instances,
an SMS alert is sent out to large numbers of people at once.
This means that if two people are scheduled to receive the
same SMS alert, they should receive them at about the same
time. SMS alerts that contain personal information, such as
banking transactions or requests for payment, are not usually
handled this way. Sending an SMS alert is often viewed by the
sender as a service. In many cases, the senders do not charge
the receivers for such messages. There may, however, be a fee
charged to both the sender and the receiver by their cellular
companies. In other cases, an SMS alert can be part of a
subscription. This is a service where a person pays a fee to
receive certain types of notifications. These can include news,
sports, and weather updates.
E. User Awareness Programs
User or Customer is the key of any field. We can develop
number of software or technology by which we can secure the
things but these all software are waist if the end user is not
getting the proper information regarding to these software. In
Banking sector there are new gadgets and technologies are
coming day by day by which the bank can provide secure
transactions to end user. Using these gadgets the banks also
Vol. 2, No.10, 2011

89 | P a g e
has to run some awareness programs for the end users by
whom they can also understand the meaning of secure
transactions as well as the user will able to learn how to use
these gadgets for secure transactions.
III. STATUS OF INDIAN BANKS WEBSITES
In this paper we take five Indian banks and try to find out
the security features using by the bank for online transactions.
The data is collected by various reports from web, newspaper
and media. For every security feature we provide 5 points. The
banks are-
State Bank of India (SBI)
Punjab National Bank [PNB]
Central Bank of India [CBI]
Bank of Baroda [BOB]
Allahabad Bank
TABLE 1. POINT TABLE
Bank PE* VK* SSL* SMS* UAP* Total
SBI 4 4 4 3 0 15
PNB 4 4 3 4 0 15
CBI 3 4 3 2 0 12
BOB 4 4 3 4 0 15
Allahabad 3 4 4 3 0 14
*PE- Password Encryption, *VK- Virtual Keyboard, *SSL-
Secure Socket Layer, *SMS- Short message service alerts,
*UAP- User Awareness Program

Figure 1. Graphical representation of security features
In this study we found that the all banks having the same
feature for his websites. They all are providing password
encryption facilities with virtual keyboard. The websites of the
banks are is using secure socket layer and they also providing
SMS alerts facilities to customer to know the information
regarding to the money transaction. We provide 5 marks for
each feature but no bank got full 5 mark for any feature and
the aggregate total of every bank is vary 12 to 15 out of 25.
The reason behind this is that they all have little bit loop
wholes on all the security features but the biggest reason is
User awareness feature. Sometime SMS alerts also come with
viruses, so this is the responsibility of the bank to check either
SMS alert is secure or not. Not all the bank user using the
SMS alert facility because the user still in confusing mode
because of the lack of awareness. One another thing is that the
most of the facilities started by the bank is still not using by
the customer because they have no proper information
regarding these facilities and if the customer is using few
facilities then this information he came to know from the other
customer. So in this area all banks have to focus properly. No
bank is providing special training programme to aware the
customer, who is the neediest person of world. Because many
of the problems user face in bank sectors due to lack of his
knowledge. So this is the responsibility of the bank to provide
the proper information to end user. They have to tell the user
What is the meaning of using virtual keyboard
What is the meaning of strong password
What is the meaning of SMS alerts
Dont access net banking account from cyber caf or
public computer.
Use a single computer as far as possible.
Login net banking site by directly typing site name.
Dont click any link, if that link takes you to login
page, close the page, and start over.
Bank or its representative never asks for password
and username over telephone.
Change the password after 6 months.
Remember the id and password, dont write it
anywhere.
Dont give any of the personal information to any
web site that does not use encryption or other secure
methods to protect it.
Dont share any information to any one regarding to
account
Install good antivirus programme on the system and
regularly update the programme.
The need for user awareness program is continuous, in
addition to being multi-disciplinary and multi-dimensional. It
is imperative to first digest that information security is a
process, not a product. An information security awareness
training program thus, needs to maintain the equilibrium
between usability, productivity and security. This paper
describes the security feature using by the bank for money
transaction on websites and focus what banks have to do for
spreading awareness. Interesting thing is that it is very difficult
to stop any kind of cyber-crime especially when we talk about
crimes related to the banks because many of the problems
occur in bank sector by the user due to lack of awareness. So if
the banks will start too aware the user then definitely the
scenario will change. We cant stop the crime, we have to face
this. The only one thing which we can do is prevention. We
have to learn how to prevent society by this kind of smart
crime and banks sector has to play key role from the front to
spread awareness.
IV. CONCLUSION AND FUTURE SCOPE
In this paper we describe different safety features using by
the banks for online transaction and examine where the
problem is in the system. We found that all the banks use the
latest technology for the online security feature but still they
Vol. 2, No.10, 2011

90 | P a g e
have small loop wholes in this feature. As well as they dont
have any user awareness program to spread information and
this is one of the biggest reasons of this online security. All
bank users do not use online facilities because they have no
proper information and the reason behind all this is the same
lack of awareness. Along the same line we people also have to
increase our awareness level because this is not only the
responsibility of the banks. One the interesting thing is that in
future these technologies will increase rapidly. It means user
will have to use these facilities therefore we need to make our
system more secure regarding to the safety mechanism. We
have to be aware of the technologies which we are using and
also need to increase our awareness level to secure
humankind.
REFERENCES
[1] Granville Williams.
[2] Proprietary Articles Trade Association v. A.G. for Canada (1932)
[3] Duggal Pawan- Cyber Crime
[4] Tan, Koontorm Center. "Phishing and Spamming via IM (SPIM)".
Retrieved December 5, 2006.
[5] Microsoft Corporation. "What is social engineering?" Retrieved August
22, 2007.
[6] "Spam Slayer: Do You Speak Spam?". PCWorld.com. Retrieved August
16, 2006
[7] "Phishing, n. OED Online, March 2006, Oxford University
Press.". Oxford English Dictionary Online. Retrieved August 9, 2006
[8] "Phishing". Language Log, September 22, 2004. Retrieved August 9,
2006.
[9] Gonsalves, Antone (April 25, 2006). "Phishers Snare Victims with
VoIP". Techweb.
[10] "Identity thieves take advantage of VoIP". Silicon.com. March 21, 2005.
[11] EP application 546704 Thomas H. Speeter/AT&T: "Intelligent work
surfaces" priority date 13.12.1991
[12] DE application 19734511 B. Kmmerer, C, Maggioni, H.
Rttger/SIEMENS AG: "Kommunikationseinrichtung" filing date
08.08.1997
[13] WO 0003348 C. Maggioni, B. Kmmerer/SIEMENS AG: "Projection
Device / Vorrichtung zur Projektion" priority date 10.07.1998
[14] EP 0554492 Hans E. Korth: "Method and device for optical input of
commands or data" filing date 07.02.1992
[15] T. Dierks, E. Rescorla (August 2008). "The Transport Layer Security
(TLS) Protocol, Version 1.2"
[16] GSM Doc 28/85 "Services and Facilities to be provided in the GSM
System" rev2, June 1985
[17] LA Times: Why text messages are limited to 160 characters
[18] GSM 03.40 Technical realization of the Short Message Service (SMS)
AUTHORS PROFILE
Susheel Chandra Bhatt holds Post Graduate degree and
pursuing Ph.D. in Computer Science. Presently, he is
working as Research Scientist in Uttarakhand Open
University, Campus Dehradun. He has five years of
teaching experience. His areas of interest are Cyber
security, computer network and database.

Prof. Durgesh Pant holds Post Graduate and Doctoral
degree in Computer Science from BIT, Mesra, India.
Presently He is working as Director, School of Computer
Science & IT, Uttarakhand Open University, Haldwani
Campus Dehradun, India. He has published more than 50
National and International papers in peer reviewed
journals, and 3 books of his credit. 12 students have been
completed their Ph.D. degree under his supervision. He is
the member of various National and International
academic, social and cultural associations and bodies.

Vol. 2, No. 10, 2011

91 | P a g e
An Empirical Study of the Applications of Web
Mining Techniques in Health Care

Dr. Varun Kumar
Department of Computer Science & Engineering
ITM University,
Gurgaon, India

MD. Ezaz Ahmed
Department of Computer Science & Engineering
ITM University,
Gurgaon, India

Abstract-Few years ago, the information flow in health care field
was relatively simple and the application of technology was
limited. However, as we progress into a more integrated world
where technology has become an integral part of the business
processes, the process of transfer of information has become
more complicated. There has already been a long standing
tradition for computer-based decision support, dealing with
complex problems in medicine such as diagnosing disease,
managerial decisions and assisting in the prescription of
appropriate treatment. Today, one of the biggest challenges that
health care system, face is the explosive growth of data, use this
data to improve the quality of managerial decisions. Web mining
and Data mining techniques are analytical tools that can be used
to extract meaningful knowledge from large data sets. This
paper addresses the applications of web mining and data mining
in health care management system to extract useful information
from the huge data sets and providing analytical tool to view and
use this information for decision making processes by taking real
life examples. Further we propose the IDSS model for the health
care so that exact and accurate decision can be taken for the
removal of a particular disease.
Keywords- Web mining; Health care management system; Data
mining; Knowledge discovery; Classification; Association rules;
Prediction; Outlier analysis, IDSS.
I. INTRODUCTION
In modern world a huge amount of data is available which
can be used effectively to produce vital information. The
information achieved can be used in the field of Medical
science, Agriculture, Business and so on. As huge amount of
data is being collected and stored in the databases, traditional
statistical techniques and database management tools are no
longer adequate for analyzing this huge amount of data.
Particular disease in a particular area and it will be related
with the agriculture of the particular area so that we can
predict prone of particular disease in a particular area due to
excessive use of pesticides and fertilizer in a particular
agricultural area. We have to collect clinical data for a
particular disease and with the help of data mining and web
mining we can predict a pattern.
Web mining as well as Data Mining (sometimes called
data or knowledge discovery) has become the area of growing
significance because it helps in analyzing data from different
perspectives and summarizing it into useful information. There
are increasing research interests in using web mining and data
mining in health care or web based health care management.
This new emerging field, called health care Web mining,
concerns with developing methods that discover knowledge
from data come from medical environments [1].
The data can be collected from various medical institutes,
doctors clinic or hospital that resides in their databases. The
data can be personal or institutional which can be used to
understand patients behavior, to assist doctor, to improve
diagnosis, to evaluate and improve health care systems , to
improve error free or zero error treatment and many other
benefits.[1][2]
Health care data mining used many techniques such as
decision trees, neural networks, k-nearest neighbor, naive
bayes, support vector machines and many others. Using these
methods many kinds of knowledge can be discovered such as
association rules, classifications and clustering. The
discovered knowledge can be used for organization of
syllabus, to predict how many patients will register for a
particular disease, alienating traditional clinical model
predicting a particular disease in a particular area.
This paper is organized as follows: Section 1 introduction
and describes the data mining techniques adopted. Section 2
discusses the application areas of these techniques in a
medical institute or clinic or any hospital. Section 3 concludes
the paper. Simply stated, data mining refers to extracting or
mining" knowledge from large amounts of data. [5]
A. Data mining techniques
Data mining techniques are used to operate on large
volumes of data to discover hidden patterns and relationships
helpful in decision making. The steps identified in extracting
knowledge from data are:
Figure1. The steps of extracting knowledge from data
Data cleaning & Integration

Data Selection & information

Data Mining & Web Mining

Pattern Evaluation

Knowledge

Data Source

Vol. 2, No. 10, 2011

92 | P a g e
B. Association analysis
Association analysis is the discovery of association rules
showing attribute-value conditions that occur frequently
together in a given set of data. Association analysis is widely
used for market basket or transaction data analysis.
More formally, association rules are of the form
" " ., . , n j m i B B A A e i Y X

, where Ai
(for i to m) and Bj (j to n) are attribute-value pairs. The
association rule X=>Y is interpreted as database tuples that
satisfy the conditions in X are also likely to satisfy the
conditions in Y ".
C. Classification and Prediction
Classification is the processing of finding a set of models
(or functions) which describe and distinguish data classes or
concepts, for the purposes of being able to use the model to
predict the class of objects whose class label is unknown. The
derived model may be represented in various forms, such as
classification (IF-THEN) rules, decision trees, mathematical
formulae, or neural networks. Classification can be used for
predicting the class label of data objects. However, in many
applications, one may like to predict some missing or
unavailable data values rather than class labels. This is usually
the case when the predicted values are numerical data, and is
often specifically referred to as prediction.
IF-THEN rules are specified as IF condition THEN
conclusion
e.g. IF age=old and patient=diabetic then heart disease
prone=yes
D. Clustering Analysis
Unlike classification and predication, which analyze class-
labeled data objects, clustering analyzes data objects without
consulting a known class label. In general, the class labels are
not present in the training data simply because they are not
known to begin with. Clustering can be used to generate such
labels. The objects are clustered or grouped based on the
principle of maximizing the intra-class similarity and
minimizing the interclass similarity.
That is, clusters of objects are formed so that objects
within a cluster have high similarity in comparison to one
another, but are very dissimilar to objects in other clusters.
Each cluster that is formed can be viewed as a class of objects,
from which rules can be derived. [5]Application of clustering
in medical can help medical institutes group individual
patient into classes of similar behavior. Partition the patient
into clusters, so those patients within a cluster (e.g. healthy)
are similar to each other while dissimilar to patient in other
clusters (e.g. disease prone or Weak).
E. Outlier Analysis
A database may contain data objects that do not comply
with the general behavior of the data and are called outliers.
The analysis of these outliers may help in disease detection
and predicting abnormal values or reason of disease.

Figure 2 Picture showing the partition of patients in clusters
II. POTENTIAL APPLICATIONS
To design, plan, build To use, interact
And maintenance participate and
Communicate

Doctor Patient

Clinical usage data
Individual data
Other data etc.

To show discovered To show recommendations
Knowledge
Figure 3 The cycle of applying web mining or data mining in health care
system
The figure No. 3 illustrates how the data from the
traditional clinic and web based health care systems can be
used to extract knowledge by applying web mining and data
mining techniques which further helps the doctors and patients
to make decisions.
A. Organization of DATA
Its important for medical institutes to maintain a high
quality healthcare program. This will improve the patients
cure process. This will also use for the optimization of
resources.
Presently, complication of disease is influenced by many
factors such as life style, other disease, environment fertilizer,
pesticides, genetic history, availability of best doctor, expert
team of doctors and experiences.
One of the applications of data mining is to identify related
disease in period of institutional programmes in a large
healthcare institute.
A case study has been performed where the patient data
collected over a period of time at healthcare Institute. The
main of the study was to find the strongly related disease in a
period offered by the institute. For this purpose following
methodology was followed:
1) Identify the possible related disease.
2) Determine the strength of their relationships and
determine strongly related disease.

Data mining and web
mining

(Clustering, classification,
outlier association, pattern
matching, text: mining.)
Healthcare systems
(Traditional clinical
healthcare system, adaptive
and intelligent web-based
health care system)
Vol. 2, No. 10, 2011

93 | P a g e
In the first step, association rule mining was used to
identify possibly related two disease combinations in the
period which also reduces our search space. In the second step,
Pearson Correlation Coefficient was applied to determine the
strength of the relationships of disease combinations identified
in the first step. [4]
TABLE 1 SHOWS THE PATIENT SUFFERING FROM THE DISEASE
Association Rules that can be derived from Table 1 are of
the form:
) 2 , ( ) 1 , ( disease X disease X

) 3 , ( ) 2 , ( ) 1 , ( disease X disease X disease X

) " " , ( ) " " , ( ure Bloodpress X Bloodsugar X [support=2%
and confidence=60%]
) " " , ( " " , ( ) " " , ( se Heartdisea X ure bloodpress X Bloodsugar X
[support=1% and confidence=50%]

Where support factor of the association rule shows that 1%
of the patient suffering from the disease blood sugar and blood
pressure, confidence factor shows that there is a chance that
50% of the patients who have Blood sugar will also have
Blood pressure
This way we can find the strongly related disease and can
optimize the database of a healthcare programme.
B. Predicting the Registration of Patients in an Hospital
Now healthcare organizations are getting strong
competition from itself, the need of better health by the
government plans and by the people concern. It needs deep
and enough knowledge for a better assessment, diagnosis,
planning, and decision making.
Data Mining helps hospitals to identify the hidden patterns
in databases; the extracted patterns are then used to build data
mining models, and hence can be used to predict diagnosis and
decision making with high accuracy. As a result of this
hospitals are able to allocate resources more effectively. [6]
One of the applications of data mining can be for example,
a hospital can take necessary actions before patient quit their
treatment, or to efficiently assign resources with an accurate
estimate of how many male or female will register for a
particular disease by using the Prediction techniques.
In real scenario couple of other associated attributes like
type of disease, hygiene environment, climate etc. can be used
to predict the registration of patients in a particular disease.
C. Predicting Patients Response Against Medicine
One of the questions, whose answer almost every doctor or
patient of a hospital would like to know Can we predict
patient performance against the medicine? [6]

Year No. of
male
patients
No. of
female
patients
2005 50 30
2006 60 35
2007 55 45
2008 65 55
2009 70 50
2010 80 ?
Figure 4 Prediction of female patient in the coming year
Over the years, many researchers applied various data
mining techniques to answer this question.
In modern times, healthcare system is taking on a more
important role in the development of our civilization by
providing good health. Good health maintenance is an
individual behavior as well as a social phenomenon.
It is a difficult task to deeply investigate and successfully
develop models for evaluating healthcare system efforts with
the combination of modernization of equipments and practice.
Healthcare organizations goals and outcomes clearly relate to
promoting good health through effective healthcare models
and research in service to healthcare system. With the help of
data mining techniques a decision making system can be
developed which can help doctors and patients to know the
weak points of the traditional clinical healthcare model. Also it
will help them to face the rapidly developing real-life
environment and adapt the current healthcare realities (such as
IDSS etc.)
We use patients response to medicine participation data as
part of the grading policy. A doctor can assess the condition of
patient by conducting an online discussion among a group of
patients and use the possible indicators such as the time
difference between medicine taken and time of response to the
medicine etc.[6] With the help of this data, we can apply
classification algorithms to classify the patients into possible
levels of grading.
D. Identifying Abnormal/ Erroneous Values
The data stored in a database may reflect outliers-|noise,
exceptional cases, or incomplete data objects. These objects
may confuse the analysis process, causing over fitting of the
data to the knowledge model constructed. As a result, the
accuracy of the discovered patterns can be poor. [5]
One of the applications of Outlier Analysis can be to detect
the abnormal values in the response sheet of the patient. This
may be due many factors like a software fault, data entry
operator negligence or an extraordinary response of the patient
for a particular disease.

Patient
id
Disease 1 Disease 2 Disease 3
1 Blood Sugar Blood Pressure Heart Disease
4 Blood Sugar Blood Pressure Kidney Problem
5 Blood Sugar Blood Pressure Eye Problem
Predictor
Unseen Data
Female
patient=65
Vol. 2, No. 10, 2011

94 | P a g e

>5 <=5

=1 >1

Figure 5 The Decision Tree built from the data in Table 2
TABLE 2 PATIENT RESPONSES DATA AND THEIR GRADES
An outlier is an observation that is numerically distant
from the rest of the data. Outliers, being the most extreme
observations, may include the sample maximum or sample
minimum or both, depending on whether they are extremely
high or low. However, the sample maximum and minimum are
not always outliers because they may not be unusually far
from other observations.
TABLE 3 THE RESPONSE OF PATIENT IN FOUR DISEASES
In the table shown above the response of the patient in
disease 4 heart strokes with patient ID 104 will be detected as
an exceptional case and can be further analyzed for the cause.
III. CONCLUSION
The focus of the study was to discuss the various data
mining techniques which can support healthcare system via
generating strategic information.
Since the application of data mining brings a lot of
advantages in higher well equipped hospitals, it is
recommended to apply these techniques in the areas like
optimization of resources, prediction of disease of a patient in
the hospital, the disease response of the patient, number of
them respond, number of the cured and number of them which
are fully satisfied.
REFERENCES
[1] C. Romero, S. Ventura, E. Garcia, "Datamining in course management
systems: Moodle case study and tutorial", Computers & Education, Vol.
51, No. 1, pp. 368-384, 2008
[2] C. Romero, S. Ventura "Educational dataMining: A Survey from 1995 to
2005", Expert Systems with Applications (33), pp. 135-146, 2007
[3] Shaeela Ayesha, Tasleem Mustafa, AhsanRazaSattar, M. Inayat Khan,
Data Mining Model for Higher Education System, Europen Journal of
Scientific Research, Vol.43, No.1, pp.24-29, 2010
[4] W. A Sandham, E.D. Lehman SIMULATING AND PREDICTING
BLOOD GLUCOSE LEVEL FOR IMPROVED DIABETES
HEALTHCARE, cmru,Imperical College, Royal Brompton London
[5] Han Jiawei, MichelineKamber, Data Mining: Concepts and Technique.
Morgan Kaufmann Publishers,2000
[6] A. Hasman, R. Bindels and P. de Clercq On the use of reminder
Systems in Healthcare, Department of Medical Informatics, University
of Maastricht, Netherlands. IEEE International Conference
[7] Sri Lanka Institute of Information Technology, http://www.sliit.lk/
[8] Sun Hongjie, Research on Student Learning Result System based on
Data Mining, IJCSNS International Journal of Computer Science and
Network Security, Vol.10, No. 4, April 2010
[9] Academy Connection Training Resources Inhtml,
http://www.cisco.com/web/learning/netacad/index, December 28th,
2005.
[10] Wayne Smith, "Applying Data Mining to Scheduling Courses at a
University", Communications of the Association for Information
Systems, Vol. 16, Article 23.
[11] Kumar, V. (2011). An Empirical Study of the Applications of Data
Mining Techniques in Higher Education. IJACSA - International Journal
of Advanced Computer Science and Applications, 2(3), 80-84.
[12] Thakur, M. (2011). Query based Personalization in Semantic Web
Mining. IJACSA - International Journal of Advanced Computer Science
and Applications, 2(2), 117-123.
AUTHORS PROFILE
Prof. Varun Kumar

Ph.D. (Computer Science), Associate Head , CSE Deptt.
School of Engineering and Technology, ITM University,
Gurgaon Haryana ,India. Presently 3 Ph. D students are
working under his supervision. Dr. Varun Kumar, completed
his PhD in Computer Science. He received his M. Phil. in
Computer Science and M. Tech. in Information Technology. He has 13 years
of teaching experience. He is recipient of Gold Medal at his Masters degree.
His area of interest includes Data Warehousing, Data Mining, and Object
Oriented Languages like C++, JAVA, C# etc. He has published more than 35
research papers in Journals/Conferences/Seminars at international/national
levels. He is working as an Editorial Board Member / Reviewer of various
International Journals and Conferences. He has 3 books, 5 study materials and
3 lab manuals to his credit.

Md. Ezaz Ahmed
Pursuing Ph.D. under the supervision of Prof, Varun kumar.
Currently, he is working with itm University as Asst.
Professor. He did his M.E (CSE) in first division with honors.
Before joining this Institute, He has more than 17 years of
experience out of which 15 years teaching and 2 years
industry experience. He has published 11 research papers, 1 in international
Journal others in national conference and in departmental journal. His area of
interest includes Web Development, software Engineering, Software
verification validation and testing, Soft Computing, and Basics of computer
and C programming. He is a member of Indian Society of Technical
Education (ISTE). Co-author of one project book published in 1998.
Time difference
between medicine
taken action (in
min)
Duration between
medicine taken and
response (in min)
Grade of the
Patients
3 1 A
3 2 B
4 1 A
5 2 B
6 1 C
6 2 C
Patient
ID
Level Blood
Sugar
ECG
Reading
Blood
pressure
reading
Heart
stroke
chance
FF PP
101 130 231 35 145/89 30
102 167 270 75 128/90 67
103 189 310 90 178/100 77
104 230 450 35 190/105 99
Time difference of
medicine action
C
Duration of medicine
taken and response
A
B
Vol. 2, No. 10, 2011

95 | P a g e
Quality EContent Design using Reusability approach

Senthil Kumar.J,
(Research Scholar, Vels University, India.) Asst.
Professor, Dhanraj Baid Jain College, India
Dr S.K.Srivatsa, Ph.D.,
Professor, St.Josephs College of Engg. , India

Abstract Technology is the one changing ever, and major
technological innovations can make a paradigm shifts. The
computer network known as the Internet is one such innovation.
After affecting sweeping changes in the way people communicate
and do business, the Internet is perched to bring about a
paradigm shift in the way people learn. Consequently, a major
change may also be coming in the way educational materials are
designed, developed, and delivered to the learner. An
instructional technology called learning objects currently leads
other candidates for the position of technology of choice in the
next generation of instructional design, development, and
delivery. This paper aims to address the reusability, generating
capability, adaptability, and scalability of the content designed
using the learning objects. Object-orientation highly values the
creation of components (called objects) that can be reused in
multiple contexts. This is the fundamental idea behind learning
objects.
Keywords- Technology; paradigm; network; learning objects.
I. INTRODUCTION
The instructional technology communities have begun to
grapple with mapping sound instructional principles to the
technical attributes of learning object systems for education and
training purposes [1]
Learning object systems are flexible, dynamic and highly
engaging technology-based environments. These systems have
great potential to capitalize on the goal-oriented nature of
human learning processes as well as allowing learners to
associate instructional content with their prior knowledge and
individual experiences
The transition of the education and training communities to
paperless, digital work and learning environments has
important implications. They often involve the re-production of
media and approaches that have been developed. Previously,
tending to increase dramatically both the cost and the time
required to develop training and education products and
services. Simply re-hosting existing education and training
approaches using digital media may optimize neither human
nor technologys capabilities.
II. INSTRUCTIONAL DESIGN BASED ON OBJECT ORIENTED
APPROACH-PEDAGOGICAL SHIFT
Before the industrial revolution, a craft based approach to
product manufacture was prevalent, where one or two
individuals create a completed product form the raw material
available to them. After the industrial revaluation there were
many changes in product manufacturing. The major
developments were the division of labor, increased automation
and to development of the component based approach to
manufacturing. The main benefit of a component based
approach is reusability is a component used on product can be
used to provide the same function for another product. Parallel
to the industrial revolution has occurred within a shorter time
frame in the software industry. It is only since the development
of the idea of software engineering in the 1970s that software
development has begun to move from a craft to an industry.
The Idea of moving to a component model for development
of courses and content has gained prominence move recently
and been driven by the interest in the educational potential of
the internet. There are number of initiatives which transfer the
ideas and benefits of the component approach to developments
and delivery of educational systems [2].
A learning component more commonly referred as a
learning object is any discrete unit of learning material that can
be extracted from one course and integrated into another.
Reusable learning object is an emerging paradigm shift in
instructional system that promises to bring to education the
same improvements in productivity that it has in software
development. There are number of problems to be resolved
before component manufactures becomes an established
approach in educational system design. These include the
issues of standards for learning objects and support for those
educators making the transition to object based design.
III. OBJECT ORIENTED APPROACH FOR CREATING
REUSABLE AND TRANSPORTABLE LEARNING CONTENT
a) Transportable among applications and environments
b) Re- purposable to different delivery structures.
To be reusable and transportable an object needs to meet
some technical coding standards and it must be instructionally
designed for reuse. In addition each learning object must be
labeled to make identification of content, topic, purpose, etc.
readily apparent and to make the object easily retrievable.
There are two requisite components of a learning object: the
object content and its metadata tag. [3] Meta-tagging means
linking or tagging objects and assets with specific metadata.
Metadata, literally data about data, is descriptive information
about a resourcemetadata allow you to locate an item very
quickly without investigating all the individual items through
which you are searching.[4] Because they are stored in a
database structure and managed through a Learning Content
Management System via meta-tagging, learning objects make it
easy to find and access content anywhere and anytime and they
are easy to update and display.
Vol. 2, No. 10, 2011

96 | P a g e
IV. FEATURES OF LEARNING OBJECTS
The following is a list of some of the types of information
that may be included in a learning object and its metadata:
General Course Descriptive Data, including: course
identifiers, language of content (English, Spanish, etc.),
subject area (Moths, Reading, etc.), descriptive text,
descriptive keywords
Life Cycle, including: version, status
Instructional Content, including: text, web pages, images,
sound, video
Glossary of Terms, including: terms, definition, acronyms
Quizzes and Assessments, including: questions, answers
Relationships to Other Courses, including prerequisite
courses
Educational Level, including: grade level, age range,
typical learning time, and difficulty.
V. METADATA STANDARDS FOR LEARNING OBJECTS
Learning objects are indeed a good idea, but as long as they
lack instructional value, we will be unable to use them
effectively. From a practical and technical perspective,
common metadata standards define what data needs to be
collected and stored to provide descriptive information about a
content object. The result is a content object metadata
specification (e.g., showing title, author, and description for
each object). Metadata standards theoretically should also
enable the appropriate use of a content object as a learning
object. In this case, the purpose is to enable learners to use one
or more learning objects to achieve one or more instructional
objectives.
The metadata on a library catalog card provides information
commonly used for finding a book or other media form, but has
little instructional information concerning the readers
instructional use of the item. If our sole purpose is to provide
metadata for describing content objects, the descriptive
information commonly included by most standards today is
sufficient. However, learning objects have important
embedded instructional objectives and, if we are not providing
instructional information in metadata, all we have is a content
object. If we ignore key instructional issues, how can we
successfully use learning objects for learning?
Many groups are working together to define common
international standards that the world can adopt for describing
learning objects that can be interoperable, reusable,
repurposable, and effectively managed and presented. Their
common interest is to find a minimum set of metadata
standards that will support the worldwide deployment of
learning objects for multiple purposes. Just a few of the groups
participating in these worldwide standards-making efforts
through the IEEE Learning Technology Standards Committee
[5] are:
Alliance of Remote Instructional Authoring and
Distribution Networks for Europe (ARIADNE, 2000)
Instructional Management Systems (IMS, 2000a) Project
Dublin Core Education Working Group (DC-Ed, 2000)
Advanced Distributed Learning Initiative (ADL, 2000)
VI. MUTABILITGY OF LEARNING OBJECT
A mutated learning object is, according to Michael
Shaw, a learning object that has been "re-purposed and/or re-
engineered, changed or simply re-used in some way different
from its original intended design". Shaw also introduces the
term "contextual learning object", to describe a learning object
that has been "designed to have specific meaning and purpose
to an intended learner". [6]
VII. PORTABILITY OF LEARNING OBJECT
Before any institution invests a great deal of time and
energy into building high-quality e-learning content (which can
cost over $10,000 per classroom hour), it needs to consider
how this content can be easily loaded into a Learning
Management System. It is possible for example, to package
learning objects with SCORM specification and load it
at Moodle Learning Management System. If all of the
properties of a course can be precisely defined in a common
format, the content can be serialized into a standard format
such as XML and loaded into other systems. When you
consider that some e-learning course need to include video
mathematical equations using MathML, chemistry equations
using CML and other complex structures the issues become
very complex, especially if the systems needs to understand
and validate each structure and then place it correctly in a
database.
VIII. LEARNING OBJECT PROJECTS
Some examples of learning object projects include:
AGORA, a publicly accessible online learning
environment at the Virtual Museum of Canada. Content is
created and produced by Canadian museum educators.
eduSource, a Canada-wide project to create the
infrastructure for a network of inter-operable learning
object repositories. The eduSource project is based on
national and international standards; it is bilingual (French
and English) and designed to be fully accessible.
MERLOT (Multimedia Educational Resource for
Learning and Online Teaching), a free and open resource
designed primarily for faculty and students of higher
education.
IQity Reactor is a learning object repository that allows
educators to create and share custom curriculum,
organized by state educational standards. Reactor is
integrated with a learning management system.
Wisc-Online is a web-based repository of learning
objects, developed primarily by faculty members from
the Wisconsin Technical College System.
IX. CONCLUSION
Object-orientation highly values the creation of components
(called objects) that can be reused in multiple contexts. This
is the fundamental idea behind learning objects. Instructional
designers can build small instructional components that can be
reused a number of times in different learning contexts.
Additionally, learning objects are generally understood to be
digital entities deliverable over the Internet, meaning that any
number of people can access and use them simultaneously.
Vol. 2, No. 10, 2011

97 | P a g e
Moreover, those who incorporate learning objects can
collaborate on and benefit immediately from new versions.
These are significant differences between learning objects and
other instructional media that have existed previously.
REFERENCES
[1] Merrill, M. D. (1999a). Instructional transaction theory (ITT):
Instructional design based on knowledge objects. In C. M. Reigeluth
(Ed.), Instructional-Design Theories and Models:
[2] Roschelle, J., Kaput, J. Stroup, W. and Kahn, T.M.., Scalable
integration of educational software: exploring the promise of component
architectures: Journal of Interactive media in education, volume 6, 1998,
www-jime.open.ac.uk/98/6
[3] Longmire, Warren. (2000). A Primer on Learning Objects.
[4] Wiley David A (2000).Connecting learning objects to Instructional
Design Theory
[5] Learning Technology Standards Committee (2002) (PDF), Draft
Standard for Learning Object Metadata. IEEE Standard 1484.12.1, New
York: Institute of Electrical and Electronics Engineers, retrieved 2008-
0429
[6] http://www.shawmultimedia.com/edtech_oct_03.html.
Vol. 2, No. 10, 2011

98 | P a g e
Retrieval of Images Using DCT and DCT Wavelet
Over Image Blocks

H. B. kekre
Professor Department of Computer Engineering
MPSTME,
NMIMS University,
Vileparle (W), Mumbai, India
Kavita Sonawane
Ph. D Research Scholar Department of Computer
Engineering MPSTME,
NMIMS University,
Vileparle (W), Mumbai, India

Abstract This paper introduces a new CBIR system based on
two different approaches in order to achieve the retrieval
efficiency and accuracy. Color and texture information is
extracted and used in this work to form the feature vector. To do
the texture feature extraction this system uses DCT and DCT
Wavelet transform to generate the feature vectors of the query
and database images. Color information extraction process
includes separation of image into R, G and B planes. Further
each plane is divided into 4 blocks and for each block row mean
vectors are calculated. DCT and DCT wavelet is applied over row
mean vector of each block separately and 4 sets of DCT and DCT
wavelet coefficients are obtained respectively. Out of these few
coefficients are selected from each block and arranged in
consecutive order to form the feature vector of the image.
Variable size feature vectors are formed by changing the no of
coefficients selected from each row vector. Total 18 different sets
are obtained by changing the no of coefficients selected from each
block. These two different feature databases obtained using DCT
and DCT wavelet are then tested using 100 query images from 10
different categories. Euclidean distance is used as similarity
measure to compare the image features. Euclidean distance
calculated is sorted into ascending order and cluster of first 100
images is selected to count the images which are relevant to the
query image. Results are further refined using second level
thresholding which uses three criteria which can be applied to
first level results. Results obtained are showing the better
performance by DCT wavelet as compare to DCT transform.
Keywords-component; DCT; DCT wavelet; Eucidean distance.
I. INTRODUCTION
Large amount of images are being generated, stored and
used daily in various real life applications through various
fields like engineering, medical sciences, biometrics,
architectural designs and drawings and many other areas.
Although various techniques are being designed and used to
store the images efficiently, still it demands to search new
effective and accurate techniques to retrieve these images
easily from large volume of databases. Text based image
retrieval techniques have tried in this direction which has got
many constraints and drawbacks associated with it which is
continuously encouraging the researchers to come up with the
new techniques to retrieve the images based on contents
instead of text annotations. Image contents are broadly
classified into global and local contents. Local contents define
the local attributes of the image like color, shape and texture
information. These attributes can be used and processed to
represent the image feature to make them comparable for
similarity. Many techniques are being developed in this field
to retrieve the images from large volume of database more
precisely [1], [2], [3], [11], [12], [13] [32] [33]. This paper
contributes in same direction by introducing the novel
techniques which are giving favorable performance which is
analyzed through different aspects of the behavior of the
proposed CBIR system.
In this work many variations are introduced which are not
used in the previous work in the same direction. We are
focusing on color and texture information of image. First we
are separating the image into R, G, B planes and then
decomposing the image plane into 4 blocks and applying DCT
transform over row mean vectors of each block of it to obtain
the texture information of the image. The logic behind that
DCT is a good approximation of principal component
extraction, which helps to process and highlight the signal
frequency features [21], [24], [26], [27], [29], [31]. Same
process is repeated with DCT wavelet transform over row
mean vectors of each block of each plane. As Wavelets can be
combined, using a "shift, multiply and sum" technique called
convolution, with portions of an unknown signal to extract
information from the unknown signal. They have advantages
over traditional fourier methods in analyzing physical
situations where the signal contains discontinuities and sharp
spikes [10], [22], [23], [28]. This paper is organized as
follows. Section II will introduce transforms applied to form
the feature vectors. Section III gives the algorithmic flow of
the system that explains how to extract the image contents and
formation of the feature vector databases [4], [5], [6], [16].
Section IV explains the experimental results with performance
analysis of the system and Section V delineate the conclusion
of the work done.
II. DISCRETE COSINE TRANSFORM AND DCT WAVELET
Discrete cosine transform is made up of cosine functions
taken over half the interval and dividing this interval into N
equal parts and sampling each function at the center of these
parts [8], the DCT matrix is formed by arranging these
sequences row wise. This paper uses DCT transform to
generate the feature vectors which is explained in section III.
Vol. 2, No. 10, 2011

99 | P a g e
Wavelets are mathematical functions that cut up the data or
signal into different frequency components by providing a way
to do a time frequency analysis. Analysis of the signals
containing the discontinuities and sharp spikes is possible with
help of wavelet transforms [7], [10], [17]. Kekres generalized
algorithm which generates the wavelet from any orthogonal
transform is used to generate DCT wavelet as DCT is an
orthogonal transform [10], [15]. To take advantage of this
property of wavelet, this paper has proposed a new algorithm
to represent the feature vectors in the form of discrete cosine
wavelet transform coefficients for the CBIR.
The DCT definition of 2D sequence of Length N is given in
equation (1) using which the DCT matrix is generated [15]
[24]. The generalized algorithm which can generate wavelet
transform of size N
2
xN
2
from any orthogonal transform of size
NxN is applied to DCT matrix and DCT Wavelet is developed
which satisfies the condition of orthogonal transforms given in
equation (2). Once the Discrete Cosine Transform Wavelet is
generated following steps are followed to form the feature
vectors of the images.
[ ] [ ]()() [
( )
[
( )
]
(1)
Where ()
{

(2)
Orthogonal: DCT Wavelet transform is said to be
orthogonal if the following condition is satisfied.
[DCTW][DCTW]
T
= [D] (3)
III. ALGORITHMIC VIEW OF CBIR USING DCT AND DCT
WAVELET
In following algorithm step1 to step3 is same for both the
approaches of CBIR
Step1. Separate the image into R, G and B planes.

Step2. Divide each plane of image into four blocks B1, B2, B3
and B4 of all equal sizes. [35]

Step3. For each block calculate the row mean vectors.
122 168 145
188
..
199 220 .. 160

Step4. In First approach we Apply Discrete Cosine Transform
over all row mean vectors of each block of each plane of the all
the database images and DCT feature database is prepared [35].
Similarly, for second approach we applied DCT wavelet over
all row mean vectors of all four blocks of each plane of all
database images and new DCT Wavelet feature database is
prepared for the second approach.
Step5. Representation of feature vectors for both the approaches
is explained as follows:
Select few DCT and DCT wavelet coefficients from each
row vector of all four blocks of each plane and arrange them in
single vector in consecutive order. It gives the feature vector
of that particular plane. Similar procedure is followed to get
the feature vector for all three planes R, G, B.
This feature vector consist of four components for each
plane for example red plane these components are named as
RB1, RB2, RB3 and RB4 where suppose each component has
64 coefficients. Arrangement of these four components in
single row vector gives the final feature vector for red plane of
size 64 x 4= 256 coefficients.
This CBIR system is experimented with various different
size feature vectors for both the approaches. Details of how
the coefficients are selected are given in following manner.

DCT and DCT Wavelet Feature vectors in Variable Size
No .of Coefficients Selected
From Each Block
1, 2, 4, 8, 12, 16, 20, 24, 28, 32, 36, 40, 44, 48,
52, 56, 60, 64
Total coefficients in the
Final Feature Vector in
Feature Database DCT and
DCT wavelet
4, 8, 16, 32, 48, 64, 80, 96, 112, 128, 144, 160,
176, 192, 208, 224, 240, 256

Feature vectors for red, green and blue plane are obtained
using above procedure and two feature vector databases are
created for all the database images using DCT and DCT
wavelet.
Step6. Once the feature databases are prepared system is tested
with query image. Feature extraction of query image will be
done in same manner as it does for the database images.
Similarity measure Euclidean distance given in equation
(5) is applied to compare the query image with the database
images for similarity [4], [5], [6], [19] [37].

B1 B2
B3 B4
(122 +168+145) /n
(199 +220+160) /n
.
.
.
.
Vol. 2, No. 10, 2011

100 | P a g e

(5)

Step7. Retrieval results are based on the criterion of sorting the
Euclidean distances in ascending order and selecting first 100
images with respect to first 100 minimum distances from 1000
distances sorted in ascending order for all database images.
III. EXPERIMENTAL RESULTS AND DISCUSSIONS
A. Database and Query Image
Algorithms discussed above in section III is experimented
with database of 1000 images which includes 100 images from
each of the following categories; that are Flower, Sunset,
Mountain, Building, Bus, Dinosaur, Elephant, Barbie, Mickey
and Horse images. Feature vectors for all these 1000 images
are extracted using above procedure based on DCT and DCT
wavelet transforms. This CBIR system is tested with query by
example image. Whenever system receives the query image it
extracts the feature vector for it in the same way as it extracts
for database images. By means of similarity measure
Euclidean distance, it will compare the query with database
images for the exact match. Ten queries from each of the 10
classes are given as query to the proposed algorithms and
Euclidean distance is calculated for all of them. Sample
Images from all classes are shown in Figure.1
B. Retrieval of Similar Images from Database of 1000 Images
Once the Query is entered it is processed as explained
above to extract its contents to form the feature vector. As
given in step 1 in section III that each image is separated into
R, G, and B planes, we are having 3 sets of feature databases
for each approach that is features for R plane, G plane and
Blue Plane. Query image along with this 3 features R, G and
B plane features will be compared with R, G and B plane
features of all database images respectively. This gives us the
3 sets of retrieval results with respect to each plane [9], [14],
[15], [20]. During the experiments of this system some
variation are made in the selection of coefficients to form the
feature vector. When we work in transform domain to utilize
and analyze the energy compaction property of them we have
selected the starting few coefficients which are carrying most
of the information of the image to represent the feature vector.
Here we have tried different size feature vectors by changing
the no of coefficients [36]. First we took all coefficients and
then we went on reducing their count to reduce the size of the
feature vector. Total 18 different sets we tried with the range
of feature vector size from 256 to 4 coefficients for each plane
and each approach.
One more variation we made in the coefficients is while
selecting the first coefficient we have scaled down it to the
range of its succeeding coefficients in that list. Because the
first coefficient is high energy coefficient as compare to all
successive coefficients. Two different scale down factors 10
and 5 are selected to just scale down the first coefficient of
each sequence. Based on these two factors, two sets of feature
databases are obtained per plane. Total 3x2x18 feature vectors
are obtained, 3 planes 2 Scale down Factors and 18 different
sizes. In turn 108 x 2 executions are made for both the
approaches for 1 query image and 216 results obtained for that
query. Like this 100 queries are tried for both the approaches
based on DCT and DCT wavelet. Table I shows the average
values of 100 query images from 10 different classes. Each
value in table is representing the average out of 10,000.
.

Figure 1. Sample Images from 10 different classes
Further, to reduce these results obtained in Table-I we have
combined the results obtained separately for each plane using
following criteria.
Criterion 1: Image will take into final retrieval only if it is
being reprieved in result set of all 3 planes R, G and B.
Criterion 2: Image will be retrieved into final retrieval only if
it is being retrieved in at least any 2 of the three planes R, G
and B.
Criterion 3: Image will be retrieved into final retrieval if it is
being retrieved in at least one of the three planes R, G and B.
All Criteria are repeated with 2 factors (10 and 5) for 100
query images. And total 3x2x100 results are obtained for each
of the two approaches based on DCT and DCT wavelet. Each
value in Table II is representing the average number of similar
images retrieved out of 10,000.
TABLE I. AVERAGE VALUES OF 100 QUERIRES FOR EACH OF THE 18SETS OF
VARIABLE COEFFICIENTS.
Retrieval Results for DCT and DCWT for Two
Scale Down Factors
Scale Down Factor 10 Scale Down Factor 5
No. of
Coefficie
-nts
DCT DCWT DCT DCWT
4 8726 8728 8740 8744
8 7270 7359 8713 8612
Vol. 2, No. 10, 2011

101 | P a g e
Retrieval Results for DCT and DCWT for Two
Scale Down Factors
Scale Down Factor 10 Scale Down Factor 5
No. of
Coefficie
-nts
DCT DCWT DCT DCWT
16 8123 8115 9350 9331
32 7938 7987 9139 9150
48 7888 7958 9065 9113
64 7686 7823 8754 8964
80 7851 7800 9006 8936
96 7831 7846 8988 9002
112 7828 7852 8993 8989
128 7828 7837 8985 8992
144 7829 7834 8983 8988
160 7685 7826 8983 8813
176 7923 7917 8979 8975
192 7644 7811 8745 8979
208 7813 7820 8973 8982
224 7810 7814 8977 8981
240 7824 7758 8976 8912
256 7812 7815 8976 8912
Observation: Scale down factor 5 gives far better performance as compare to factor 10 and DCWT
results are better than DCT.
C. Results and analysis of CBIR using DCT and DCT wavelet.
Proposed algorithm is experimented with 100 queries, 10
images from each category and results are obtained by
applying the similarity measure Euclidean distance. Retrieval
results of all 18 sets of feature vector of different sizes are
obtained for each plane separately along with two scale down
factors 10 and 5. Table I is showing the average values for
execution of 100 queries for each feature vector set from 18
variations. When we observed these results obtained using
the scale down factor 5 are giving best performance in all the
sets and for all the planes R, G, and B. It can also be noticed in
the chart 1and 2 that performance of factor 5 is having good
accuracy as compare to factor 10.
When we observed the these results obtained for each one
18 sets of different size feature vectors for all 3 planes it has
been noticed that for first few coefficients sets selection
system is performing well. It has been observed that when
feature vector size was 16 coefficients for factor 5 and 4
coefficients for factor 10, system has given its best
performance as shown in Chart 1 and 2. One more
observation made that red plane is proving its best in terms of
the average values of retrieval set. To refine the results
obtained for 3 planes we have applied the above mentioned 3
criteria and the results obtained are shown in Table II for the
both the approaches with reference to both factors 5 and 10.
In these results we can notice that criterion 3 is giving best
performance among all three sets of results where the image
similar to query will be retrieved in final set if it is being
retrieved in at least one of three planes. Chart 3, 4, and Chart
5, 6 are displaying the results for all criteria for DCT and
DCWT for factor 10 and 5 respectively, where we can notice
the behavioral difference of the system for these 3 criteria as
mentioned above.
D. Performsance Evaluation of CBIR using DCT and DCT
wavelet.
Results obtained in this work using DCT and DCT
Wavelet, is indirectly compared with the traditional
parameters Precision and Recall. Here when system generates
the retrieval result in terms of 1000 Euclidean distances
between the given query image and 1000 database images
which are sorted in ascending order; out of which first 100
images are selected as retrieval set of similar images which
carries images belong to same category of query and even
other category images as well [18]. When we talk in terms of
precision, it is in the range of 30% to 70% for most of the
query images. At the same time very good results are obtained
for most of the query images for both the approaches in terms
of recall parameter which is in the range of 40% to 90 % for
many query images.
We compare these results with the other work done using
DCT or other wavelets [15], [29], [30], [35]. It can be
observed and noticed that the database we are using includes
images from different classes and each class has got 100
images of its own category which has got images with
different background also which has impact on the feature
extraction and even on the retrieval process. It still performs
better in terms of precision and recall. We have tried 100
query images and the cumulative result which is average of
100 queries is summarized in the above tables. If we consider
the result of each query separately in most of the queries can
say for 50 % of the queries we have got very good values for
precision which is around 0.7 to 0 .8. and at same time for the
same query are getting good results in terms of recall which is
around 0.6 to 0.7 which can be considered as best results for
CBIR system. But at the same time if we have to consider the
overall performance of these approaches they should perform
well or in same manner for all 100 or more queries which
again triggering us to make future improvements. This is
explained in brief in the last section after conclusion.

Vol. 2, No. 10, 2011

102 | P a g e
TABLE II. AVERAGE VALUES OF 100 QUERIRES FOR EACH OF THE 18SETS OF VARIABLE COEFFICIENTS FOR ALL 3 CRITERIA FOR FACTOR 10
DCT with Scale down factor 10 DCWT with Scale down factor 10
Size of Feature
vector Criterion1 Criterion2 Criterion3 Criterion1 Criterion2 Criterion3
4 1293 2653 4900 1294 2653 4901
8 941 2199 4284 833 2076 4410
16 1237 2522 4412 1228 2535 4419
32 1216 2477 4298 1238 2493 4318
48 1233 2456 4233 1248 2473 4285
64 1152 2395 4203 1154 2367 4346
80 1245 2468 4200 1156 2372 4329
96 1244 2458 4193 1244 2464 4197
112 1242 2455 4187 1245 2464 4199
128 1244 2461 4189 1245 2564 4188
144 1244 2457 4191 1247 2461 4185
160 1248 2452 4181 1245 2388 4206
176 1248 2452 4174 1245 2455 4178
192 1090 2344 4264 1251 2453 4169
208 1248 2451 4167 1250 2454 4171
224 1249 2450 4169 1253 2454 4171
240 1246 2455 4170 1163 2360 4290
256 1249 2454 4160 1247 2453 4162
TABLE III. AVERAGE VALUES OF 100 QUERIRES FOR EACH OF THE 18SETS OF VARIABLE COEFFICIENTS FOR ALL 3 CRITERIA FOR FACTOR 5
DCT with Scale down factor 5 DCWT with Scale down factor 5
Size of Feature
vector Criterion1 Criterion2 Criterion3 Criterion1 Criterion2 Criterion3
4 1285 2648 4902 1286 2646 4898
8 1272 2692 4817 1138 2586 4957
16 1569 2946 4889 1574 2951 4888
32 1527 2861 4768 1555 2901 4820
48 1546 2852 4716 1548 2871 4762
64 1430 2754 4636 1438 2760 4834
80 1547 2760 4687 1441 2838 4817
96 1551 2839 4678 1550 2839 4667
112 1555 2836 4660 1553 2837 4669
128 1553 2837 4659 1556 2836 4662
144 1554 2839 4653 1553 2844 4658
160 1553 2839 4652 1558 2775 4683
176 1553 2839 4648 1558 2836 4650
192 1364 2716 4741 1558 2834 4653
208 1556 2837 4649 1557 2834 4650
224 1554 2836 4653 1555 2836 4655
240 1561 2838 4652 1449 2758 4766
256 1560 2832 4652 1559 2833 4644

Observation: DCWT with scale down factor 5 gives better performance under all three criteria. No of images retrieved increases from criterion 1 to criterion 3
Vol. 2, No. 10, 2011

103 | P a g e

Figure 2. Plot for DCT and DCWT Using Scale Down Factor 10 for

Figure 3. Plot for DCT and DCWT using Scale Down Factor 5

Figure 4. Plot for All 3 Criteria Using Scale Down Factor 10 for DCWT
DCT
DCWT
I
m
a
g
e
s

r
e
t
r
i
e
v
e
d

o
u
t

o
f

1
0
,
0
0
0

Size of Feature Vector
Criterion1
Criterion2
Criterion3
I
m
a
g
e
s

r
e
t
r
i
e
v
e
d

o
u
t

o
f

1
0
,
0
0
0

Vol. 2, No. 10, 2011

104 | P a g e

Figure 5. Plot for All 3 Criteria Using Scale Down Factor 10 for DCT

Figure 6. Plot for All 3 Criteria Using Scale Down Factor 5 for DCWT

Figure 7. Plot for All 3 Criteria Using Scale Down Factor 5 for DCT

IV. CONCLUSION
CBIR system based on DCT and DCWT has been studied
through many different aspects of its behavior in this paper. It
mainly focuses on application of two transforms DCT and
DCWT, their performance analysis and comparative study.
This includes many things within it. In both the algorithms
each image is divided into 3 planes that mean color
information is handled separately to form the feature vectors.
As each plane is divided into 4 blocks and transforms are
applied to row mean vectors of each block which tells that the
texture of the image is taken into consideration while forming
the feature vectors.
By changing the size of the feature vectors using 18
different sets computational time complexity is analyzed and it
can be defined that computational time can be saved with
smaller size feature vectors which are performing better as
compared to the larger ones.
Criterion1
Criterion2
Criterion3
I
m
a
g
e
s

r
e
t
r
i
e
v
e
d

o
u
t

o
f

1
0
,
0
0
0

Criterion1
Criterion2
Criterion3
I
m
a
g
e
s

r
e
t
r
i
e
v
e
d

o
u
t

o
f

1
0
,
0
0
0

Criterion1
Criterion2
Criterion3
I
m
a
g
e
s

r
e
t
r
i
e
v
e
d

o
u
t

o
f

1
0
,
0
0
0

Vol. 2, No. 10, 2011

105 | P a g e
Along with the different size of feature vectors Systems
performance is also checked using the scale down factors 10
and 5 which actually stabilizes the high energy of first DCT or
DCWT coefficient, brings it into the same range of remaining
low energy coefficients. This gives the strong improvement in
the retrieval results as shown in chart 1 and 2.
As three planes are handled separately each time 3 results
sets are obtained which are further combined using three
criteria to prove the best out of it where we can notice that
criteria 3 is giving best performance among all 3.
Finally when we compare DCT and DCWT it can be
noticed that DCWT is performing better. The best
performance is given by DCWT with factor 5 at 16
coefficients as shown in figure 2. If properties of wavelet
taken into consideration we can say that all small details of the
image can be extracted to form the feature vectors and also
maximum computational time can be saved as compare to
normal DCT transform. Required multiplications using DCT
for N x N blocks are N
2
where DCWT requires only N (2N-1)
multiplications which saves considerable computational time
of the system and gives better performance as well.
On the basis of comparison of this work with existing
systems many places we found our results are better in terms
of similarity retrieval and also in terms of computational time
required. But as there is scope for further improvement so that
these approaches can be used for variable image sizes and
along with color and texture shape feature can also be
considered for the comparisons and also the overall average
precision and recall can further be improved for all 100 or
more queries towards the ideal value.
REFERENCES
[1] Remco C. Veltkamp, mirela tanase department of computing
science,utrecht university, content-based image retrieval systems: a
survey Revised and extended version of technical report uu-cs- 2000-
34,october october 28, 2002.
[2] Yixin chen, member IEEE, james z. Wang, member IEEE, and
robertkrovetz clue: Cluster-Based Retrieval Of Images By
UnsupervisedLearning IEEE Transactions On Image Processing,
Vol.14, No. 8,August 2005.
[3] Qasim Iqbal And J. K. Aggarwal, Cires: A System For Content-
BasedRetrieval In Digital Image Libraries Seventh International
ConferenceOn Control, Automation, Robotics And Vision (Icarcv02),
Dec 2002,Singapore.
[4] Guoping Qiu Color Image Indexing Using Btc IEEE Transactions On
Image Processing, Vol. 12, No. 1, January 2003.
[5] R. W. Picard and T. P. Minka, Vision texture for annotation, J.
Multimedia Syst., vol. 3, no. 1, pp. 314, 1995.
[6] S. Santini and R. Jain, Similarity measures, IEEE Trans. Pattern
Anal.Mach. Intell., vol., 2005, in press.
[7] E. de Ves, A. Ruedin, D. Acevedo, X. Benavent, and L. Seijas, A New
Wavelet-Based Texture Descriptor forImage Retrieval, CAIP 2007,
LNCS 4673, pp. 895902, 2007, Springer-Verlag Berlin Heidelberg
2007.
[8] H.B.Kekre, Dhirendra Mishra, DCT-DST Plane sectorization of
Rowwise Transformed color Images in CBIR International Journal of
Engineering Science and Technology, Vol. 2 (12), 2010, 7234-7244.
[9] H. B. Kekre, Kavita Patil, Standard Deviation of Mean and Variance of
Rows and Columns of Images for CBIR, International Journal of
Computer and Information Engineering 3:1 2009.
[10] H. B. Kekre, Archana Athawale, Dipali Sadavarti Algorithm to
Generate Wavelet Transform from an Orthogonal Transform,
International Journal Of Image Processing (IJIP), Volume (4): Issue (4).
[11] H.B.Kekre, Sudeep D. Thepade, Priyadarshini Mukherjee, Shobhit
Wadhwa, Miti Kakaiya, Satyajit Singh, Image Retrieval with Shape
Features Extracted using Gradient Operators and Slope Magnitude
Technique with BTC, International Journal of Computer Applications,
September 2010 issue. (0975 8887) Volume 7 No.10, October 2010.
[12] Samy Ait-Aoudia1, Ramdane Mahiou1, Billel Benzaid, Yet Another
Content Based Image Retrieval system, 1550-6037/10 $26.00 2010
IEEE, DOI 10.1109/IV.2010.83
[13] P. S. Hiremath , Jagadeesh Pujari, Content Based Image Retrieval using
Color, Texture and Shape features, -7695-3059-1/07$25.00 2007
IEEE, 10.1109/ADCOM.2007.21.
[14] H. B. Kekre Kavita Sonawane, CBIR Using Kekres Transform over
Row column Mean and Variance Vectors, International Journal on
Computer Science and Engineering,Vol. 02, No. 05, 2010, 1609-1614.
[15] H. B. Kekre, Kavita Patil, DCT over Color Distribution of Rows and
Columns of Image for CBIR Sanshodhan A Technical Magazine of
SFIT No. 4 pp. 45-51, Dec.2008.
[16] H.B.Kekre, Sudeep D. Thepade, Image Retrieval using Augmented
Block Truncation Coding Techniques, ACM International Conference
on Advances in Computing, Communication and Control
(ICAC3-2009), pp.: 384-390, 23-24 Jan 2009,Fr. Conceicao Rodrigous
College of Engg., Mumbai. Available online at ACM portal.
[17] H.B.Kekre, Dhirendra Mishra, Performance comparison of four, eight
and twelve Walsh transform sectors feature vectors for image retrieval
from image databases, Iinternational journal of Engineering, science
and technology(IJEST) Vol.2(5) 2010, 1370-1374 ISSN 0975-5462.
[18] H.B.Kekre, Tanuja Sarode, Sudeep D. Thepade, Color-Texture Feature
based Image Retrieval using DCT applied on KekresMedian
Codebook, International Journal on Imaging (IJI), Volume 2, Number
A09, Autumn 2009,pp. 55-65. Available online at
www.ceser.res.in/iji.html (ISSN: 0974-0627).
[19] Subrahmanyam Murala, Anil Balaji Gonde, R. P. Maheshwari, Color
and Texture Features for Image Indexing and Retrieval, 2009 IEEE
International Advance Computing Conference (IACC 2009), Patiala,
India, 6-7 March 2009.
[20] H. B. Kekre Ms. Kavita Sonawane, Feature Extraction in Bins Using
Global and Local thresholding of Images for CBIR, Published in
International Journal of Computer, Information and System Science and
Engineering. (IJCISSE, Vol. 3, No. 1, Winter 2009 pp.1- 4).
[21] Lu, Z.-M.; Burkhardt, H., Colour image retrieval based on DCT domain
vector quantization index histograms Electronics Letters Volume 41,
Issue 17, 18 Aug. 2005 pp: 956 957.
[22] Combes J, Grossmann A, Tchamitchian P. Wavelets: Time-Frequency
Methods and Phase Space. 2. Springer-Verlag; 1989.
[23] M.K. Mandal, T. Aboulnasr, S. Panchanathan, Image indexing using
moments and wavelets, IEEE Trans. Consumer Electron. Vol. 42 No
3,1996,pp.557-565.
[24] Yung-Gi Wu1, Je-Hung Liu, Image Indexing in DCT Domain ,
Proceedings of the Third International Conference on Information
Technology and Applications (ICITA05)
[25] C.C Chang, J.C Chuang, Y.S Hu, Retrieving digital images from a
JPEG compressed image database, Digital Image Vision Computing
Vol. 22 2004, pp. 471-484.
[26] K. Ait saadi, A.Zemouri, Z. Brahimi , H.Meraoubi. Indexing and
Retrieval Medical images based on 2X2 DCT and IDS Compression,
Proceedings of the 2005 5th International Conference on Intelligent
Systems Design and Applications (ISDA05).
[27] Ramesh Babu Durai, V.Duraisamy A generic approach to content
based image retrieval using dct and classification techniques, (IJCSE)
International Journal on Computer Science and Engineering Vol. 02, No.
06, 2010, 2022-2024.
[28] H.B.Kekre, Tanuja Sarode, Sudeep D. Thepade, Image Retrieval using
Color-Texture Features from DCT on VQ Codevectors obtained by

Vol. 2, No. 10, 2011

106 | P a g e
Kekres Fast Codebook Generation, ICGST-GVIP Journal, Volume 9,
Issue 5, September 2009, ISSN: 1687-398X.
[29] M.Babu Rao, B.Prabhakara Rao, A.Govardhan, Content based image
retrieval using Dominant color and Texture features, International
Journal of Computer science and information security,Vol.9 issue No:2,
February 2011.pp:41-46.
[30] Kishore Kumar et al. Content based image retrieval - extraction by
objects of user interest, International Journal on Computer Science and
Engineering (IJCSE), Vol. 3 No. 3 Mar 2011.
[31] Gwangwon Kang, Junguk Beak Features Defined by Median Filtering
on RGB Segments for Image Retrieval, Second UKSIM European
Symposium on Computer Modeling and Simulation, 978-0-7695-3325-
4/08, 2008 IEEE.
[32] Yu-Len Huang and Ruey-Feng Chang , Texture features for dct-coded
image Retrieval and classification, 0-7803-5041 -3/99, 1999 IEEE.
[33] Chong-Wah Ngo, Ting-Chuen Pong, Exploiting image indexing
techniques in DCT domain, Pattern Recognition 34 (2001) 1841-1851
Published by Elsevier Science Ltd.
[34] S.Cheng, W. Huang, Y. Liao and D. Wu, A Parallel CBIR
Implementation Using Perceptual Grouping Of Block-based Visual
Patterns, IEEE International Conference on Image Processing ICIP,
2007, pp. V -161 - V 164,
[35] Mann-Jung Hsiao, Yo-Ping Huang, Te-Wei Chiang, A Region-Based
Image Retrieval Approach Using Block DCT, 0-7695-2882-1/07,
2007 IEEE
[36] Kekre Transform over Row Mean, Column Mean and Both Using Image
Tiling for Image Retrieval, International Journal of Computer and
Electrical Engineering, Vol.2, No.6, December, 2010, 1793-8163.
[37] M. Saadatmand-Tarzjan and H. A. Moghaddam, A Novel Evolutionary
Approach for Optimizing Content-Based Image Indexing Algorithms,
IEEE Transactions On Systems, Man, And CyberneticsPart B:
Cybernetics, Vol. 37, No. 1, February 2007, pp. 139-153.

AUTHORS PROFILE
Dr. H. B. Kekre has received B.E. (Hons.) in
Telecomm. Engg. from Jabalpur University in
1958, M.Tech (Industrial Electronics) from IIT
Bombay in 1960, M.S. Engg. (Electrical Engg.)
from University of Ottawa in 1965 and Ph.D.
(System Identification) from IIT Bombay in
1970. He has worked Over 35 years as Faculty
of Electrical Engineering and then HOD
Computer Science and Engg. at IIT Bombay. For last 13 years
worked as a Professor in Department of Computer Engg. at Thadomal
Shahani Engineering College, Mumbai. He is currently Senior
Professor working with Mukesh Patel School of Technology
Management and Engineering, SVKMs NMIMS University, Vile
Parle(w), Mumbai, INDIA. He has guided 17 Ph.D.s, 150
M.E./M.Tech Projects and several B.E./B.Tech Projects. His areas of
interest are Digital Signal processing, Image Processing and
Computer Networks. He has more than 350 papers in National /
International Conferences / Journals to his credit. Recently twelve
students working under his guidance have received best paper
awards. Five of his students have been awarded Ph. D. of NMIMS
University. Currently he is guiding eight Ph.D. students. He is
member of ISTE and IETE.
Ms. Kavita V. Sonawane has received M.E
(Computer Engineering) degree from Mumbai
University in 2008, currently Pursuing Ph.D. from
Mukesh Patel School of Technology, Management
and Engg, SVKMs NMIMS University, Vile-Parle
(w), Mumbai, INDIA. She has more than 8 years of
experience in teaching. Currently working as a Assistant professor in
Department of Computer Engineering at St. Francis Institute of
Technology Mumbai. Her area of interest is Image Processing, Data
structures and Computer Architecture. She has 7 papers in National/
International conferences / Journals to her credit.

Vol. 2, No. 10, 2011

107 | P a g e
Cryptanalysis of an Advanced Authentication Scheme

Sattar J Aboud
Department of Information Technology
Iraqi Council of Representatives
Baghdad-Iraq
Abid T. Al Ajeeli
Department of Information Technology
Iraqi Council of Representatives
Baghdad-Iraq

AbstractIn this paper we study a scheme for making
cryptanalysis and security improvement. This protocol by Song,
is a password authentication protocol using smart card. We note
that this protocol has been shown to be prone to the offline
password guessing attack. We perform an additional
cryptanalysis on this scheme and detect that it is vulnerable to the
clogging attack, a type of denial-of-service attack. We notice that
all smart card typed authentication schemes which lead the
scheme by Song, and need the server to find the computationally
exhaustive modular exponentiation, similar to the scheme by Xu
et al., and it is vulnerable to the clogging attack. Then we propose
an enhancement in the scheme to avoid the clogging attack.
Keywords-authentication protocol; offline password guessing
attack; clogging attack.
I. INTRODUCTION
The idea behind improving password authentication
protocol is to help authorized users obtain services from an
authorized server. When an entity needs a service from a
server, it has to identify itself to the server in a certain way.
Password authentication has been one of the most suitable
techniques for a user ID throughout the years. Currently,
millions of providers utilize password authentication schemes
to identify authorized users. General cases include private web
service, Internet shopping, e-mail service, e-trade service, and
other services. Fundamentally, each password authentication
method has two steps:
- Registration step: In this step, the user enters a user ID and
password in the computer. The password is saved by the
server and kept confidential between the entity and the
computer.
- Authentication step: In this step, the entity needs a service
from the computer. It passes its identity and password to
the computer to get a service. The computer then decides if
the user is authorized by checking the received information
of user ID password with the saved details. The server
extends the preferred service to the entity, if found
authorized
In authentication processes, the password is sent cross an
insecure communication channel. A hacker can intercept the
message by listening to the communication. It can imitate the
entity by re-using the password acquired from the
communication. These issues compromise the entity
confidentiality. The service providers normally keep
passwords of users in a database for potential verification and
authentication. The passwords are kept in a password index in
the computer database. It does not provide any security against
unprivileged insiders of the computer, and it does not protect
the passwords when the computer database is someway
hacked. To reduce the problem of password list disclosure, the
computer can encrypt the passwords and protect them.
Nevertheless, the communication interception remains a threat
to the organization security. Another problem is recalling a
user identity and password. A compromise in the password for
any entity can be like losing a credit card. The hacker can take
advantage before the corporation is informed about the loss.
Therefore we have to improve efficient password
authentication protocol to finish the authentication process in a
secured way.
Taking this into consideration, and to produce a more
secure scheme, various smart cards typed password
authentication schemes have been proposed throughout the
last decade [4, 6, 7]. In such a system, the entity is given with
a smart card. When the user needs a service, it gives its smart
card with a password that remains private. The smart card then
employs this password to build a login message that is passed
to the computer. The computer then authenticates this message
and gives the preferred service if the password is found
legitimate.
In this paper, we study such a scheme under smart card by
Song [1]. Song presented the scheme as an enhancement of
another scheme set is by Xu et al. [3]. The scheme has been
shown to be prone to the imitation attack. Song then
considered this competitive scheme in the same study, to avoid
the imitation attack on the scheme. In this paper, we will show
that the Song protocol is prone to the clogging attack, a type of
the denial-of-service-attack. We noted that the scheme in [3]
was vulnerable to clogging attack and also prone to clogging
attack. In this attack, the hacker can easily prevent the
computer from giving any service without having any
information related to the identity or password of the entity.
The hacker wants to make any complex computations to
launch a clogging attack on Song scheme. We have come to
know a cryptanalysis of the Song scheme by Tapiador et al
[2]. However, they did not study the clogging attack in the
scheme in their research. It also worth mentioning that the
chain of smart card typed authentication schemes that keeps
the server computationally exhaustive modular exponentiation
[5] are all prone to the clogging attack.
The hacker takes the benefit of the calculation
intensiveness of the modular exponentiation computation in
initiating this attack. To avoid clogging attack on these
schemes key agreement schemes are required in different
Vol. 2, No. 10, 2011

108 | P a g e
types of communication. Keys want to be securely exchanged
before a channel can be recognized. There are few security
threats to this that are intruder-in-the-middle so that the hacker
pretends to be someone else than connecting participants.
Replay of old keys is one more attack which is common in this
viewpoint. Therefore it is necessary to improve secure key
exchanging schemes to create secure channel. The key
agreement protocols commonly have two steps:
- Determine the public and private keys: In this step, both
the participants compute a pair of keys: the secret key,
which is kept private, and the public key, which is made
public. In some schemes, a part of this step is completed by
a key distribution center that keeps the public keys of the
entities in a directory.
- Determine the secret session key: In this step, the users
swap their public keys with a certain integer. The secret
session key for message is computed using those integers,
and the secret and public keys. Several schemes also let a
certain participant determine the private key or session key
and pass it to the other participant by encrypting it with the
message. In this paper we will not study any scheme
regarding the key agreement protocols.
The rest of the paper is organized as follows: in section 2,
we study the Song scheme including some an enhancement we
suggest concerning public and private keys ; then we give a
toy example in section 3. Then we have a cryptanalysis of
Song scheme .including clogging attack and offline password
guessing attack in section 4. We then propose a solution to this
attack in section 5. In section 6, we conclude this work.
II. REVIEW OF SONG SCHEME
We summarize here the password authentication
protocol of Song [1]. This authentication protocol uses a smart
card. In section 3, we consider the achievement and security
vulnerability of this protocol. In section 4, we introduce a
clogging attack on the scheme. We also consider a possible
solution against the attack. Finally, we show that similar
protocols of [3], that let the server calculate modular
exponentiation, are vulnerable to the same type of attack and
has the same achievement dependencies as this protocol. The
Song protocol contains three steps: registration, login and
authentication. Prior to starting with the registration phase, the
server performs the following steps:
1. two primes p and q are chosen where 1 * 2 + = q p .
2. an integer
*
q
Z i e is chosen.
3. a hash function h is chosen.
4. an integer e is chosen as encryption key such that
1 ) 1 , gcd( = p e .
5. the decryption key d is computed, where
1 mod
1
=

p e d .
6. i and d are determined as both are private keys.

A. Registration Phase
This phase comprises the following steps:
1. Entity A selects ) , (
A A
w id
2. Entity A passes ) , (
A A
w id to the server over a secure
way, such that
A
id and
A
w are the user-id and
password of entity A respectively.
3. The server calculates ) ( ) mod (
A
i
A A
w h p id h z = .
4. The server saves the parameter ) , , , ( e h z id
A A
into a
smart card.
5. The server sends the smart card to the entity A .
B. Logical Phase
Entity A performs the following steps:
1. Provides its ) , (
A A
w id .
2. Chooses an arbitrary integer
A
r .
3. Sets
A
T system present time.
4. Finds ) (
A A A
w h z y = .
5. Computes ) (
A A y A
T r e x
A
= .
6. Computes ) | | | | | | (
A A A A A
id x r T h u = .
7. Sends the message ) , , , (
A A A A
T x u id to the server.
C. Authentication Phase
The following steps are performed by the server:
1. the server performs the following after receiving the login
request from an entity:
- Verifies that
A
id , and
A
T . If not, rejects the login
message.
- Finds ) mod ( p id h y
i
A A
= .
- Computes
A A y A
T x d r
A
= ) (
'

- Verifies that ) | | | | | | (
'
A A A A A
id x r T h u . Otherwise
rejects the login message.
- Finds ) | | | | (
'
S A A S
T r id h u = ,
- Passes the message ) , , (
S S A
T u id to an entity A .
2. The following steps are performed by an entity A after
the receipt of ) , , (
S S A
T u id from a server:
- authenticates
A
id and
S
T
- Checks that ) | | | | (
S A A S
T r id h u . If identical, the
server is validated.
3. The entity A and the server then find the session key s as
follows:
- Entity A compute the session key as follows

) | | | | | | (
A A S A
r T T id h s =
- The server computes the session key as follows
) | | | | | | (
'
A A S A
r T T id h s = .
- Finally, a session key s is agreed by the two
participants.

D. Comments
When the entity wants to alter the password
A
w to a new
password
'
A
w , the smart card will first verify the validity of
Vol. 2, No. 10, 2011

109 | P a g e
A
w through interacting with the server, and if it succeeds, it
resets
A
w to
'
A
w ,and substitutes
A
z with
' '
A A A A
w w z z = .
III. TOY EXAMPLE YOUR
Suppose that the server selects 23 = p , then 11 = q .
Suppose that 2 = i , and for simplicity ignore in this phase the
hash function (.) h . Suppose the encryption key 7 = e , and then
compute the decryption key d as follows
1 22 mod 63 * 7 1 mod 1 * = = p d e .
Registration Phase: Suppose that entity A selects
) 24 , 29 ( = =
A A
w id , then entity A passes ) 24 , 29 ( = =
A A
w id
to the server over a secure way so that
A
id and
A
w are the
user-id and password of entity A . Then, the server calculates
) ( ) mod (
A
i
A A
w h p id h z =

21 ) 24 ( ) 23 mod 29 (
2
= = h h . The server then saves the
parameter ) 7 (.), , 21 , 29 ( = = = = e h z id
A A
into a smart card.
Finally, the server sends the smart card to the entity A .
Login Phase: First, the smart card is attached to the card
reader by an entity A and it provides ) 24 , 29 ( = =
A
w id . The
smart card then chooses a random integer ) 12 ( =
A
r and then
sets ) 4 ( =
A
T . Then entity A computes
13 ) 24 ( 21 ) ( = = = h w h z y
A A A
. Then compute
12 ) 4 12 ( 13 ) ( = = =
A A y A
T r e x
A
. compute 16 23 mod 12
7
= =
A
x .
Then compute

) | | | | | | (
A A A A A
id x r T h u =

15 23 mod 61 ) 29 16 12 4 ( = = + + + = h . Then entity A sends the
message ) 4 , 16 , 15 , 29 ( = = = =
A A A A
Authentication Phase: First, the server verifies that
A
id is
valid. If not, it rejects the login message. Then,
13 ) 23 mod 29 ( ) mod (
2
= = = h p id h y
i
A A
is computed. Then
= = ) ) mod ((
'
A
d
A y A
T p x d r
A

12 23 mod 13 * 8 ) 4 12 ( ) 12 23 mod ) 16 (( 13
63
= = = = is computed.
Then, the server verifies that

) | | | | | | (
'
A A A A A
id x r T h u
15 23 mod 61 ) 29 16 12 4 ( = = + + + = h . If the verification of the
previous step succeeds, set
s
T as current server time. Suppose
that 4 =
s
T , and then compute
22 23 mod 45 ) 4 12 29 ( ) | | | | (
'
= = + + = = h T r id h u
S A A S
.
Finally, the server passes the message ) 4 , 22 , 29 ( = = =
S S A
T u id

to entity A . An entity A authenticates
A
id and
S
T by Checking
that ) | | | | (
S A A S
T r id h u 22 ) 4 12 29 ( = + + = h . In this
example, it is identical; it means that the server is validated.
However, the entity A and the server then find the session key
s as follows: first, entity A computes the session key as
follow ) | | | | | | (
A A S A
r T T id h s = 3 23 mod 49 ) 12 4 4 29 ( = = + + + = h
.Then, the server computes the session key as follows
) | | | | | | (
'
A A S A
r T T id h s = 3 23 mod 49 ) 12 4 4 29 ( = = + + + = h .
IV. SONG SCHEME CRYPAYLISIS
The clogging attack and offline password guessing attack
will be discussed in this section.
A. The Clogging Attack
This scheme has a large dependency on the computer and
entity clocks. For a connection-oriented use, this may be
unwieldy. The scheme should be designed to care for time
synchronization between clocks of different entities and
servers. This scheme should be made fault tolerant to deal
with complex network faults and also with different types of
attacks. Despite that the communication is secure, a chance of
an attack can occur and a hacker may intercept a message and
alter its timestamp
A
T . In this way the hacker successfully
repudiates the authorized entity because the server will refuse
the login message on the basis of timestamp dissimilarity.
Thus this type of attack is probable even if the scheme avoids
replay attacks. Also, interacting delays can prepare the
timestamp to go beyond the threshold thus making the entire
service inherently decelerate. Therefore, we illustrate that
Song scheme is vulnerable to the clogging attack. The
clogging attack is a type of attack by which the hacker H
constantly passes messages to a server and clogs it with those
messages [8]. Suppose this could occur with the Song
scheme. The following is done by a hacker H :
1. H intercepts the message
) 4 , 12 , 11 , 29 ( = = = =
A A A A
T x u id passed by an entity to
a server in the login phase.
2. H can alter the timestamp 4 =
A
T to some 10 =
u
T ,
because the message is unencrypted. The change
satisfies the criterion g T T
u
s
*
.
3. H alters 11 =
A
u to arbitrary nonsense value 9 =
u
u .
4. H passes ) 10 , 12 , 9 , 29 ( = = = =
u A u A
T x u id to the
server.
The following is performed by the server:
1. Verify if 29 =
A
id is valid. At this point it is valid.
2. Find ) mod ( p id h y
i
A A
= 13 ) 23 mod 29 (
2
= = h .
3. Find ) (
'
u A y A
T x d r
A
= 9 ) 10 12 ( 13 = =
4. verify ) | | | | | | (
'
A A A u u
id x r T h u =
14 23 mod 60 ) 29 12 9 10 ( = = + + + = h . This does
not succeed, thus the message is rejected.
Then, the hacker H will continue repeating the steps many
times and let the server calculate the modular exponentiation
repeatedly. Essentially, H can potentially alter all the entering
login requests from the authorized entity to the server. As
modular exponentiation is computationally exhaustive, the
victimized server spends large processing resources doing
ineffective modular exponentiation rather than any actual
work. Therefore the hacker H clog the server with ineffective
work and so repudiates any authorized entity. The hacker only
wants an id of a valid entity to achieve the clogging attack
many times.
Vol. 2, No. 10, 2011

110 | P a g e
B. Offline Password Gussing Attack
In [1] it is claimed that the hacker should not be able to
attack and get access to the server by extracting the
information kept on the smart card. But, a hacker who gets the
result of ) ( ) mod (
A
i
A A
w h p id h z = can simply increase an
offline password guessing attack by easily viewing one right
authentication session and obtaining access to the results of
A
x , and
A
u . The attack is performed as follows: for every
password
*
A
w , the hacker calculates the uncertain encryption
key ) (
* *
A A A
w h z y = . Such a key is then employed to
decrypt the nonce result
*
A
r by first recovering
A
x with
*
A
y and
then the value with
A
T (both of which are public); namely,
A A y A
T x d r
A
= ) (
* *
. Note that when an attempt of password
*
A
w
is right (i.e., )
*
A A
w w = , then it is obtained encryption key
*
A
y
and so, the nonce
*
A
r . Now,
A
u can be used to verify when
that is the case. The hacker finds ) | | | | | | (
* *
A A A A A
id x r T h u
and, when it coincides with
A
u , it can conclude that
*
A
r is
right and thus the password tried. In this reasoning we suppose
that h has no collisions. Yet, even when h is not perfect, extra
eavesdropping sessions can be employed to exclude false
positives and find the right password. Briefly, in addition to
what is claimed in [1], the messages exchanged during the
scheme certainly decrease the entropy of the password, at least
for a hacker with access to the values stored in the card. Also,
once the password is guessed, the scheme provides no
protection against other attacks.
C. Comments
It can be observed that the attack showed can also be made
on the schemes by Xu et al. [3] and by Tsaur et al. [5]. Thus
we notice that the clogging attack can be executed on all the
smart card typed authentication schemes using computing
modular exponentiation.
V. THE POSSIBLE SOLUTION
We will discuss the possible solution for the problem
raised.
A. Prevent The Clogging Attack
At the start of the authentication phase, the server will
check if the IP address of the entity is valid. It has to identify
the IP addresses of any registered authorized users. Despite
that, hacker H might spoof the IP address of an authorized
entity and replay the login request. To stop it, we may add a
cookie exchange step at the start of the login phase of Song
scheme. This step has been presented as in the familiar Oakley
key exchange scheme [9].
1. The entity A selects an arbitrary number
1
m and passes
it with the message ) , , , (
A A A A
2. The server accepts the message and passes its own
cookie
2
m to the entity A .
B. Solution Discussion
When the hacker H spoofs the entity IP address, H will
not obtain
2
m back from the server. But H just succeeds to
have the server return an acknowledgement, not to calculate
the computationally modular exponentiation. Thus the
clogging attack is prevented by these extra steps. We note that
this process does not avoid the clogging attack but only
frustrates it to a certain extent.
VI. CONCLUSION
We have studied a scheme in this paper. The scheme is a
password authentication protocol; which we have shown to be
vulnerable to the clogging attack. We demonstrated that the
attack on this scheme could be prevented by using an extra
step of exchanging numbers. We demonstrated that it is prone
to man-in-the-middle attack. Then we showed how to prevent
this attack by using an encryption and decryption algorithm.
We indicated a security get-out as Song proposed, which is
that the hacker can execute modular exponentiation on both
sides of the authentication scheme. In addition, after
intercepting the retrieved information, the hacker can start new
logon information and successfully log into the server system.
Thus, Song proposition cannot give adequate security and it is
not appropriate for practical implementation of the proposed
scheme.
REFERENCES
[1] Ronggong Song, "Advanced smart card based password authentication
protocol", Computer Standards & Interfaces, Volume 32, Issue 4, pp. 321-
325, June 2010.
[2] Juan E. Tapiador, Julio C. Hernandez-Castro, Pedro Peris-Lopez, John A.
Clark, "Cryptanalysis of Songs advanced smart card based password
authentication protocol", Unpublished manuscript, June 2010.
[3] Jing Xu,Wen-Tao Zhu, and Deng-Guo Feng, "An improved smart card
based password authentication scheme with provable security", Computer
Standards & Interfaces, Volume 31, Issue 4, pp. 723-728, June 2009.
[4] Chang. C.C., Wu T.C., "Remote password authentication scheme with
smart cards", IEEE Proceedings Computers and Digital Techniques,
Volume138, Issue 3, pp.165168, 1991.
[5] Woei-Jiunn Tsaur, Chia-Chun Wu, and Wei-Bin Lee, "A smart card based
remote scheme for password authentication in multi-server Internet
services", Computer Standards & Interfaces, Volume 27, pp. 39-51, June
2004.
[6] Wen-Shenq Juang, "Efficient password authenticated key agreement
using smart cards", Computers & Security, Volume 23, Issue 2, pp. 167-
173, March 2004.
[7] Chien H.Y., Jan J.K., and Tseng Y.M., "An efficient and practical
solution to remote authentication: smart card", Computers and Security,
vol.21, no.4, pp.372375, 2002.
[8] Tseng Y.M., "An efficient two-party identity-based key exchange
protocol", Informatica 18 (1) pp. 125136, 2007.
[9] Hsi-Chang Shih, "Cryptanalysis on Two Password Authentication
Schemes", Master Thesis, Laboratory of Cryptography and Information
Security Department of Computer Science and Information Engineering,
National Central University, Chung-Li, Taiwan 320, Republic of China,
2006.
[10] Meshram, C. (2010). Modified ID-Based Public key Cryptosystem using
Double Discrete Logarithm Problem. IJACSA - International Journal of
Advanced Computer Science and Applications, 1(6).
[11] Meshram, C. (2011). Some Modification in ID-Based Public key
Cryptosystem using IFP and DDLP. IJACSA - International Journal of
Advanced Computer Science and Applications, 2(8).
AUTHORS PROFILE
Sattar J Aboud is a Professor and advisor for Science and Technology at
Iraqi Council of Representatives. He received his education from United
Vol. 2, No. 10, 2011

111 | P a g e
Kingdom. Dr. Aboud has served his profession in many universities and he
awarded the Quality Assurance Certificate of Philadelphia University, Faculty
of Information Technology in 2002. Also, he awarded the Medal of Iraqi
Council of Representatives for his conducting the first international
conference of Iraqi Experts in 2008. His research interests include the areas of
both symmetric and asymmetric cryptography, area of verification and
validation, performance evaluation and e-payment schemes.
Abid T. Al Ajeeli is a Professor at Iraqi Council of Representatives. He
received his education from United Kingdom. Dr. Ajeeli has served his
profession in many universities and he awarded the Medal of Iraqi Council of
Representatives for his conducting the first international conference of Iraqi
Experts in 2008. His research interests include the areas of software
engineering, verification and validation, information security and simulation.

Vol. 2, No. 10, 2011

112| P a g e
Conceptual Level Design of Semi-structured
Database System: Graph-semantic Based Approach

Anirban Sarkar
Department of Computer Applications
National Institute of Technology, Durgapur
West Bengal, India

AbstractThis paper has proposed a Graph semantic
based conceptual model for semi-structured database
system, called GOOSSDM, to conceptualize the different
facets of such system in object oriented paradigm. The
model defines a set of graph based formal constructs,
variety of relationship types with participation constraints
and rich set of graphical notations to specify the
conceptual level design of semi-structured database
system. The proposed design approach facilitates modeling
of irregular, heterogeneous, hierarchical and non-
hierarchical semi-structured data at the conceptual level.
Moreover, the proposed GOOSSDM is capable to model
XML document at conceptual level with the facility of
document-centric design, ordering and disjunction
characteristic. A rule based transformation mechanism of
GOOSSDM schema into the equivalent XML Schema
Definition (XSD) also has been proposed in this paper. The
concepts of the proposed conceptual model have been
implemented using Generic Modeling Environment
(GME).
Keywords- Semi-structured Data; XML; XSD; Conceptual
Modeling; Semi-structured Data Modeling; XML Modeling.
I. INTRODUCTION
The increasingly large amount of data processing on the
web based applications has led a crucial role of semi-
structured database system. In recent days, semi-structured
data has become prevalent with the growing demand of such
web based software systems. Semi-structured data though is
organized in semantic entity but does not strictly conform to
the formal structure to strict types. Rather it possess irregular
and partial organization [1]. Further semi-structured data
evolve rapidly and thus the schema for such data is large,
dynamic, is not strict to type and also is not considered the
participation of instances very strictly.
The eXtensible Markup Language (XML) is increasingly
finding acceptance as a standard for storing and exchanging
structured and semi-structured information over internet [12].
The Document Type Definition (DTD) or XML Schema
Definition (XSD) language can be used to define the schema
which describes the syntax and structure of XML documents
[9]. However, the XML schemas provide the logical
representation of the semi-structured data and it is hard to
realize the semantic characteristics of such data. Thus it is
important to devise a conceptual representation of semi-
structured data for designing the information system based on
such data more effectively. A conceptual model of semi-
structured data deals with high level representation of the
candidate application domain in order to capture the user ideas
using rich set of semantic constructs and interrelationship
thereof. Such conceptual model will separate the intention of
designer from the implementation and also will provide a
better insight about the effective design of semi-structured
database system. The conceptual design of such system further
can be implemented in XML based logical model.
To adopt the rapidly data evolving characteristics, the
conceptual model of semi-structured data must support several
properties like, representation of irregular and heterogeneous
structure, hierarchical relations along with the non
hierarchical relationship types, cardinality, n array relation,
ordering, representation of mixed content etc. [13]. Beside
these, it is also important to realize the participation
constraints of the instances in association with some relation
or semi-structured entity type. The participation of instances in
semi-structured data model is not strict. In early years, Object
Exchange Model has been proposed to model semi-structured
data [2], where data are represented using directed labeled
graph. The schema information is maintained in the labels of
the graph and the data instances are represented using nodes.
However, the separation of the structural semantic and content
of the schema is not possible in this approach. In recent past,
several researches have been made on conceptual modeling of
semi-structured data as well as XML. Many of these
approaches [3, 4, 5, 6, 7, 8] have been extended the concepts
of Entity Relationship (ER) model to accommodate the facet
of semi-structured data at conceptual level. The major
drawbacks of these proposals are in representation of
hierarchical structure of semi-structured data. Moreover, only
two ER based proposals [7, 8] support the representation of
mixed content in conceptual schema. In [7], a two level
approach has been taken to represent the hierarchical relations.
In first level the conceptual schema is based on extended
concept of ER model and in second level, hierarchical
organizations of parts of the overall conceptual schema are
designed. In general, ER model are flat in nature [14] and thus
unable to facilitate the reuse capability and representation of
hierarchical relationship very efficiently. On the other hand,
ORA-SS [11] proposed to realize the semi-structured data at
conceptual level starting from its hierarchical structure. But
the approach does not support directly the representation of
Vol. 2, No. 10, 2011

113| P a g e
no-hierarchical relationships and mixed content in conceptual
level semi-structured data model.
Very few attempts have been made to model the semi-
structured data using Object Oriented (OO) paradigm. ORA-
SS [11] support the object oriented characteristic partially. The
approaches proposed in [9, 10, 12] are based on UML. These
approaches support object oriented paradigm comprehensively
and bridge the gap between OO software design and semi-
structured data schemata. However, the UML and extensions
to UML represent software elements using a set of language
elements with fixed implementation semantics (e.g. methods,
classes). Henceforth, the proposed approaches using extension
of UML, in general, are logically inclined towards
implementation of semi-structured database system. This may
not reflect the facet of such system with high level of
abstraction to the user. In other word, semi-structured data
model with UML extension cannot be considered as
semantically rich conceptual level model. In [16] a graph
semantic based web data model has been proposed and is
appropriate for modeling structured web database system. The
approach has not considered semi-structured characteristics of
web databases.
In this paper, a graph semantic based conceptual model for
semi-structured database system, called Graph Object
Oriented Semi-Structured Data Model (GOOSSDM), has been
proposed. The model is comprehensively based on object
oriented paradigm. Among others, the proposed model
supports the representation of hierarchical structure along with
non-hierarchical relationships, mixed content, ordering,
participation constraints etc. The proposed GOOSSDM
reveals a set of concepts to the conceptual level design phase
of semi-structured database system, which are understandable
to the users, independent of implementation issues and provide
a set of graphical constructs to facilitate the designers of such
system. The schema in GOOSSDM is organized in layered
approach to provide different level of abstraction to the users
and designers. In this approach a rule based transformation
mechanism also has been proposed to represent the equivalent
XML Schema Definitions (XSD) from GOOSSDM schemata.
The correctness of such transformation has been verified using
the structural correlation mechanism described in [15].
Moreover, the concepts of proposed GOOSSDM have been
implemented using Generic Modeling Environment (GME)
[14] which is a meta-configurable modeling environment. The
GME implementation can be used as prototype CASE tools
for modeling semi-structured databases using GOOSSDM.
The preliminary version of this work has been published in
[17] which has been now enriched and completed with
comprehensive formalization and CASE tools.
II. GOOSSDM: THE PROPOSED MODEL
The GOOSSDM extends the object oriented paradigm to
model semi-structured data. It contains all the details those are
necessary to specify the irregular and heterogeneous structure,
hierarchical and non-hierarchical relations, n array
relationships, cardinality and participation constraint of
instances. The proposed data model allows the entire semi-
structured database to be viewed as a Graph (V, E) in layered
organization. At the lowest layer, each vertex represents an
occurrence of an attribute or a data item, e.g. name, day, city
etc. Each such basic attribute is to be represented as separate
vertex. A set of vertices semantically related is grouped
together to construct an Elementary Semantic Group (ESG). So
an ESG is a set of all possible instances for a particular
attribute or data item. On next, several related ESGs are group
together to form a Contextual Semantic Group (CSG). Even the
related ESGs with non-strict participations or loosely related
ESGs are also constituent of related CSG the constructs of
related data items or attributes to represent one semi-structured
entity or object. The edges within CSG are to represent the
containment relation between different ESG in the said CSG.
The most inner layer of CSG is the construct of highest level of
abstraction or deeper level of the hierarchy in semi-structured
schema formation. This layered structure may be further
organized by combination of one or more CSGs as well as
ESGs to represent next upper level layers and to achieve further
lower level abstraction or higher level in the semi-structure
data schema hierarchy. From the topmost layer the entire
database appears to be a graph with CSGs as vertices and edges
between CSGs as the association amongst them. The CSGs of
topmost layer will act as roots of semi-structured data model
schemata.
A. Modeling Constructs in GOOSSDM
Since from the topmost layer, a set of vertices V is decided
on the basis of level of data abstraction whereas the set of
edges E is decided on basis of the association between
different semantic groups. The basic components for the
model are as follows,
A set of t distinct attributes A = {a
1
, a
2
, ., a
t
} where,
each a
i
is an attribute or a data item semantically distinct.
(a) Elementary Semantic Group (ESG): An elementary
semantic group is an encapsulation of all possible instances or
occurrences of an attribute, that can be expressed as graph
ESG (V, E), where the set of edges E is a null set and the set
of vertices V represent the set of all possible instances of an
attribute x
i
A. ESG is a construct to realize the elementary
property, parameter, kind etc. of some related concern.
Henceforth there will be set of t ESGs and can be represented
as E
G
= {ESG
1
, ESG
2
, ., ESG
t
}. The graphical notation for
the any ESG is Circle.
(b) Contextual Semantic Group (CSG): A lowest layer
contextual semantic group is an encapsulation of set of ESGs
or references of one or more related ESGs to represent the
context of one entity of semi-structured data. Let, the set of n
CSGs can be represented as C
G
= {CSG
1
, CSG
2
, , CSG
n
}.
Then any lowest layer CSG
i
C
G
can be represented as a
graph (V
Ci
, E
Ci
) where vertices V
Ci
E
G
and the set of edges
E
Ci
represents the association amongst the vertices. For any
CSG, it is also possible to designate one or more encapsulated
ESGs as determinant vertex which may determine an
unordered or ordered set of instances of constituent ESGs or
CSGs. The graphical notation for any CSG is square and
determinant vertex is Solid Circle.
Composition of multiple CSGs can be realized in two
ways. Firstly, the simple Association (Discussed in subsection
II.B) may be drawn between two or more associated CSGs
Vol. 2, No. 10, 2011

114| P a g e
either of same layer or of adjacent layers to represent the non-
hierarchical and hierarchical data structure respectively. The
associated CSG will be connected using Association
Connector. Those CSGs may share a common set of ESGs or
referred ESGs.
Secondly, lower layer CSGs may maintain an Inheritance
or Containment relationship (Discussed in subsection 2.B)
with the adjacent upper layer CSG to represent the different
level of abstraction. Thus, the upper layer CSG can be formed
by inheritance or composition of one or more lower layer
CSGs along with encapsulation of zero or more related ESGs
or reference of ESGs. Then any upper layer CSG
i
C
G
can be
represented as a graph (V
Ci
, E
Ci
) where vertices V
Ci
C
G
E
G

Reference (E
G
)

and the set of edges E
Ci
represents the
association amongst the vertices.
(c) Annotation: Annotation is a specialized form of CSG and
can be expressed as G(V, E), where |V| = 1 and E = .
Annotation can contain only text content as tagged value.
Annotation can be containment in or associated with any other
CSG. Further the cardinality constraint for Annotation
construct is always 1:1 and ordering option can be 1 or 0,
where 1 means content will be in orderly form with other
constituent ESG and 0 means text content can be mixed with
other constituent ESGs. The annotation construct will realize
the document-centric semi-structured data possibly with mixed
content. This concept is extremely important for mapping
semi-structured data model in XML. Graphically Annotation
can be expressed using Square with Folded Corner.
The summary of GOOSSDM constructs and their
graphical notations have been given in Table I.
TABLE I. SUMMARY OF GOOSSDM CONSTRUCTS AND THEIR GRAPHICAL
NOTATIONS
GOOSSDM
Constructs
Description Graphical
Notation
ESG Elementary Semantic Group

Determinant
ESG
Determinant vertex of any CSG
which will determine the other
member vertices in the CSG

CSG Contextual Semantic Group
Annotation Specialized form of CSG. Contain
only Text Content.

Association
Connector
Connect multiple associated CSGs

B. Relationship Types in GOOSSDM
The proposed GOOSSDM provides a graph structure to
represent semi-structured data. The edges of the graph
represent relationships between or within the constructs of the
model. In the proposed model, four types of edges have been
used to represent different relationships. The type of edges and
their corresponding meanings are as follows,
(a) Containment: Containments are defined between
encapsulated ESGs including determinant ESG and parent
CSG, or between two constituent CSGs and parent CSG, or
between CSG and referential constructs. The Containment
relationship is constrained by the parameters tuple <p, >,
where p determines the participation of instances in
containment and determines the ordering option of
constituent ESGs or CSGs. With any CSG, this represents a
bijective mapping between determinant ESG and other ESGs
or composed CSG with participation constraint. Graphically
association can be expressed using Solid Directed Edge from
the constituent constructs to its parent labeled by constraint
specifications. The possible values for p are as follows,
(i) 1:1 Represents ESG with mandatory one instantiation or
total participation in the relationship and is default value of p.
(ii) 0:1 Represents ESG with optional one instantiation in the
relationship.
(iii) 1:M Represents ESG with mandatory multiple
instantiation in the relationship.
(iv) 0:M Represents ESG with optional multiple
instantiation in the relationship.
(v) 0:X Represents ESGs with optional exclusive
instantiation in the relationship. If a CSG contain single such
ESG then it will act like 0:1 option. Otherwise one such ESG
will optionally instantiate among all ESGs with p value 0:X.
(vi) 1:X Represents ESG with mandatory exclusive
instantiation in the relationship. If a CSG contain single such
ESG then it will act like 1:1 value option. Otherwise it is
mandatory that one such ESG will instantiate among all ESGs
with p value 1:X.
The possible values for are as follows,
(i) 1 Represents that for any CSG, the constituent ESGs
and CSGs are ordered and order must be maintained from left
to right in the list of ESGs with value 1.
(ii) 0 This is default value of and represents that for any
CSG, the constituent ESGs and CSGs are not ordered.
(b) Association: Associations are defined between related
CSGs of same layer of adjacent layers. The Association
relationship is constrained by the parameters tuple <P, >,
where P determines the cardinality of Association and
determines the ordering option of associated CSGs.
Graphically association can be expressed using Solid
Undirected Edge. Any CSG wish to participate in association
will be connected with association relationship. On next,
multiple associated CSGs will be connected through
Association Connector. For semi-structured it is sometime
important to have specific context of some association. Such
context can be represented using Associated CSG defined on
Association Connector. Association Connector facilitates the n
array relationship. Graphically association can be expressed
using Solid Undirected Edge, Association connector can be
expressed using Solid Diamond and Associated CSG can be
connected with Association Connector using Dotted
Undirected Edges with Participation constraint specifications.
The values for P can be 1:1 or 0:1 or 1:N or 0:N or 0:X or 1:X
with corresponding meaning and possible values for can be
1 or 0 with corresponding meaning.
(c) Link: Links are used to represent the inheritance
relationships between two CSGs. Graphically link can be
expressed using Solid Directed Edge with Bold Head from the
generalized CSG to the specialized one.

Vol. 2, No. 10, 2011

115| P a g e
(d) Reference: In semi-structured data model, it is important to
represent the symmetric relationship between ESGs or CSGs.
Reference can be used to model such concepts. Reference
relations are defined either between ESG and referred ESG or
between CSG and referred CSG. Graphically reference can be
expressed using Dotted Directed Edge.
The summary of GOOSSDM relationship types and their
graphical notations have been given in Table II.
TABLE II. SUMMARY OF GOOSSDM RELATIONSHIP TYPES AND THEIR
GRAPHICAL NOTATIONS
GOOSSDM
Relationships
Description Graphical
Notation
Containment Defined between Parent CSG and
constituent ESGs and CSGs

Association Defined between CSGs of same
layer or adjacent layers.

CSG Association Defined between association and
associated CSG

Link Defined between two adjacent
layer parent CSG and inherited
CSG

Reference Defined either between ESG and
referred ESG or between CSG and
referred CSG.

III. TRANSFORMATION OF GOOSSD INTO XSD
In general, the proposed GOOSSDM can be useful to
realize the semi-structured data schema at conceptual level.
The logical structure of such schema can be represented using
the artifacts of XSD. Moreover, XSD is currently the de facto
standard for describing XML documents. An XSD schema
itself can be considered as an XML document. Elements are the
main building block of any XML document. They contain the
data and determine the elementary structures within the
document. Otherwise, XSD also may contain sub-element,
attributes, complex types, and simple types. XSD schema
elements exhibit hierarchical structure with single root element.
A systematic rule based transformation of GOOSD
schemata to XSD is essential to express the semi-structured
data at logical level more effectively. For the purpose, a set of
rules have been proposed to generate the equivalent XSD from
the semantic constructs and relationship types of a given
GOOSSDM schemata. Based on the concepts of GOOSSDM
constructs and relationship types the transformation rules are as
follows,
Rule1: An ESG will be expressed as an element in XSD. For
example, ESG
City
can be defined on attribute City to realize
Customer city. Any DESG construct must be expressed with
typed ID in XSD. The equivalent representation in XSD can be
as given in Figure 1.

Figure1. Representation of ESG
Rule 2: A CSG will be expressed as a complexType in XSD.
For example, CSG
Customer
can be defined to realize the detail of
Customer. The equivalent representation in XSD can be as
given in Figure 2.
Rule 3: Any Annotation construct will be expressed as a
complexType in XSD with suitable mixed value. On
containment to other CSG, if value is 0 then it will be treated
as mixed content in the resulted XML document. Otherwise if
value is 1 then it will be treated as annotation text in resultant
XML document in orderly form. An example of annotation
constructs has been shown in Figure 3.

Figure 2. Representation of CSG

Figure 3. Representation of Annotation Construct
Rule 4: A Reference of ESG and CSG will be expressed as a
complexType in XSD. For example, a reference of ESG
City
can
be defined on attribute City to realize a referential attribute on
Customer city. The equivalent representation in XSD can be as
given in Figure 4.

Figure 4. Representation of Reference of ESG.
Rule 5: CSGs of topmost layer will be treated as root in XSD
declaration.
Rule 6: Any lowest layer CSG with containment of some ESGs
will be expressed as a complexType with elements declaration
in XSD. Further the participation constraint (p value in
GOOSSDM concept) can be expressed using minOccurs and
maxOccurs attribute in XSD. The ordering constraint ( value
in GOOSSDM concept) can be expressed using compositor
type of XSD. If value is 1 then compositor type will be
sequence otherwise all. For ordered set ESGs, the order will be
maintained from left to right. If any subset of ESGs contains
the p value X:1 then those ESGs will be composite using
choice compositor type in XSD. An example of XSD
representation of lowest layer CSG has been shown in Figure 5.
Rule 7: Any upper layer CSG with containment of ESGs,
reference of ESGs and adjacent lower layer CSGs will be
<p, >
<P, >
<P, >

Customer
<xs:complexType name="Customer"/>

Vol. 2, No. 10, 2011

116| P a g e

Figure 5. Representation of Lower layer CSG

Figure 6. Representation of Upper layer CSG.
expressed as a complexType in XSD. An example of XSD
representation of upper layer CSG with containment relation
has been shown in Figure 6.
Rule 8: Any upper layer CSG with Link relationship with
adjacent lower layer CSGs will be expressed as a complexType
with inheritance in XSD. Upper layer CSG will be the child of
lower layer CSG. An example of XSD representation of upper
layer CSG with inheritance with adjacent lower layer CSG has
been shown in Figure 7.
Rule 9: Any upper layer CSG with Association relationship
with adjacent lower layer CSGs will be expressed as a
complexType with nesting in XSD. Upper layer CSG will be
treated as root element.
Rule 10: Association relationship between any two CSGs in the
same layer will be expressed as a complexType with nesting in
XSD. Rightmost CSG will be treated as the root element and
on next nesting should be done in right to left order of the CSG
in the same layer.
Rule 11: N array Association relationship within a set of
CSGs spread over several layer will be expressed as a
complexType with nesting in XSD. Topmost layer CSG will be
treated as the root element in XSD. Then, in the adjacent lower
layer the rightmost CSG should be treated as nested element
within the root element. Further the nesting should be done in
right to left order of the CSG in the same layer and on next
moving on the adjacent lower layers.

Figure 7. Representation of Link Relationship
Rule 12: With several Association relationships (composition
of n array and simple relationship) within a set of CSGs
spread over several layer will be expressed as a complexType
with nesting in XSD. Topmost layer CSG will be treated as the
root element in XSD. Then, if available, the directly associated
CSGs in each adjacent lower layer will be nested till it reaches
Vol. 2, No. 10, 2011

117| P a g e

Figure 8. Representation of Associated CSGs spread over several layers (ESG layer is hidden).
to the lowermost layer of available associated CSG. On next,
the CSGs of adjacent lower layer of the root element will be
nested from right to left order in the same layer along with the
nesting of directly associated CSGs (if available) in each
corresponding adjacent layers. An example has been shown in
Figure 9 for XSD representation of GOOSSDM schemata
where associated CSGs are spread over three layers and contain
both n array and simple associations.

Figure 9. Irregular Structure in Visit Records XML
IV. CASE STUDY
Let consider an example of Visit Record of Patient where a
Patient can visit to a Doctor either at Hospital Department or
at Clinic [7]. Any patient can visit several times to different
doctors. The Figure 9 shows an irregularly structured XML
representation of visit records of two patients. Patient 1 visited
twice to two different Doctors, one at Hospital Department and
another at Clinic. Patient 2 visited once to one Doctor common
to Patient 1 but at different Clinic. All though in the document,
the Date of Visit, Doctor and option of Hospital Department
and Clinic are in order. The XML document of Figure 9
represents the semi-structured data for such Visit Record
database. The suitable GOOSSDM schemata for such data and
its equivalent XSD have been shown in Figure 10. The
equivalent XSD of GOOSSDM schemata of Figure 10 can be
generated using the rules described in Section III.
V. CORRECTNESS OF GOOSSDM TRANSFORMATION
The set of proposed transformation rules described in
Section III facilitates the systematic transformation of
conceptual level semi-structured data model like GOOSSDM
to the equivalent XSD in logical level. The correctness of the
model transformation can be proved using the structural
correspondence approach described in Narayanan et al [15]. In
every model transformation, there is a correlation or
correspondence between parts of the input model and parts of
the output model. One can specify these correlations in terms
of the abstract semantics of the source and target model
constructs. The approach of Narayanan et al. describes that, if a
transformation has resulted in the desired output models, there
will be a verifiable structural correspondence between the
source and target model instances that is decidable. Moreover,
the transformation can be accepted as correct, if a node in the
Vol. 2, No. 10, 2011

118| P a g e
source model and its corresponding node in the target model
satisfy some correspondence conditions.

Figure 10. Corresponding GOOSSDM schemata and Equivalent XSD of
Figure 9

In case of GOOSSDM, the meta-model level identifiable
correspondence structures are listed in Table III and the table
can be treated as the look-up table for the cross links between
the source model (GOOSSDM) and target model (XSD).
Further, the proposed set of rules will realize the
correctness condition in model transformation. In our proposed
approach, the correspondence rules must ensure that semantics
and syntax for every constructs in the GOOSSDM model and
its instance being transformed into the XSD model elements.
Several examples have been illustrated for the proposed
transformation rules to verify the correctness of
correspondence mapping of GOOSSDM schemata to the
equivalent XSD.
TABLE III. LOOK-UP TABLE FOR STRUCTURAL CORRESPONDANCE
GOOSSDM
Constructs
Graphical
Notation
Equivalent XSD Representation
ESG xs:Element
Determinant
ESG
xs:ID
CSG xs:complexType
Annotation

xs:complexType with suitable Mixed
value
Association
Connector
xs:complexType with nesting from right
to left order
Containment xs:complexType with element
declaration
Association xs:complexType with nesting
CSG
Association
xs:complexType with nesting from right
to left order
Link xs:extension declaration
Reference ref declaration in xs:element
P value 1:1 or 0:1 or
1:N or 0:N
or 0:X or
1:X
minOccurs and maxOccurs delcarations
value 1 or 0 Compositor type : all or sequence
VI. FEATURES OF GOOSSDM
The proposed GOOSSDM is an extension of
comprehensive object oriented model for Semi-structured
Database System and which can be viewed as a Graph (V, E) in
layered organization. It contains set of semantically enriched
constructs and relationship types to describe all the details
those are necessary to specify the artifacts of the system
containing semi-structured data. Moreover, using proposed set
of rules, the proposed model schemata can be systematically
transformed into equivalent XSD, which represents the logical
schema for semi-structured data. Apart from these, one of the
major advantages of the model is that it defines each level of
structural detail on the constructs which are independent of the
implementation issues. Moreover, the graph structure maintains
the referential integrity inherently. The features of the proposed
model are as follows,

<p, >
<P, >
<P, >
Vol. 2, No. 10, 2011

119| P a g e
(a) Explicit Separation of structure and Content: The model
provides a unique design framework to specify the design for
the semi-structured database system using semantic definitions
of different levels (from elementary to composite) of data
structure through graph. The model reveals a set of structures
like ESG, CSG, Annotation, Association Connector etc. along
with a set of relationships like Containment, Association,
Link, Reference etc. between the structures, which are not
instance based or value based. So, the nature of contents that
corresponded with the instances and the functional constraint
on the instances has been separated from the systems
structural descriptions.
(b) Abstraction: In the proposed GOOSSDM, the concepts of
layers deploy the abstraction in semi-structured data schema.
The upper layer views will hide the detail structural
complexity from the users. Such a representation is highly
flexible for the user to understand the basic structure of semi-
structure database system and to formulate the alternative
design options.
(c) Reuse Potential: The proposed model is based on Object
oriented paradigm. It is supported with inheritance mechanism
using the Link relationship. Henceforth, there is no binding in
the model to reuse some CSG constructs of any layer. On
reuse of CSG, the specialized CSG must be shown in adjacent
upper layer of the parent CSG. Moreover, lowest layer ESG or
lower layer can be shared and reused with different CSGs of
the upper layers using Containment relationship.
(d) Disjunction Characteristic: The instances of semi-
structured data schema are likely to be less homogeneous than
structured data. Disjunction relationships facilitate the
possibility of non-homogeneous instances. The proposed
GOOSSDM supports disjunction relationship using the
participation constraint attribute p or P (by setting p or P value
either 0:X or 1:X). The Containment relationships between
constituent ESGs or CSGs with the parent CSG can be
disjunctive or Association relationships between two or more
CSGs can be disjunctive. Figure 5 and Figure 10 respective
explain such disjunctions.
(e) Hierarchical and Non-hierarchical Structure: The
proposed model explicitly supports both hierarchical and non-
hierarchical representation in semi-structure data modeling at
conceptual level. Associated CSGs of different or same layers
form the hierarchical or non-hierarchical structure in semi-
structured data model. At the logical level modeling of semi-
structured data using XSD supports only hierarchical structure.
For the purpose, the set of rules have been proposed to
transform more generous conceptual level schema to
hierarchical logical schema.
(f) Ordering: Ordering is one important concept in modeling
of semi-structured data. One or more attributes or relationships
in semi-structured data schema can be ordered. Our proposed
model supports ordering in two ways using the relationship
ordering constraint attribute. Firstly, the ordering may be
enforced between parent CSG and any set of constituent ESGs
and CSGs by specifying the value on containment
relationship. Secondly, the ordering can be enforced on the
any set of Association relationships within CSG.
(g) Irregular and Heterogeneous structure: By characteristic
the semi-structured data is irregular and heterogeneous. The
proposed GOOSSDM supports disjunction characteristic,
ordering and representation of both hierarchical and non-
hierarchical structure in the same schema. With all these
facets, the proposed model can efficiently model the irregular
and heterogeneous semi-structured data. Modeling of irregular
structure using GOOSSDM has been shown in Figure 10.
(h) Participation constraint: Instances participations in the
semi-structured data schema are not followed strictly.
Participations of instances can be optional or mandatory or
even exclusive for such schema. This can affect the
participation of constituent ESGs and CSGs in the parent CSG
or may affect the participation of CSGs in some association
relationship either of simply type or n-array type. All these
participation constraint can be modeled in proposed
GOOSSDM by specifying the value for participation
constraint attribute p or P.
(i) Document-centric and Mixed Content: In real world,
document texts are mixed with semi-structured data. The
feature is more important and frequent in XML documents.
Thus it is an essence that, the conceptual model for semi-
structured data must support modeling of such feature. In the
proposed model, the Annotation construct facilitates to model
document centric design of semi-structured data at conceptual
level. Moreover, the modeling of the Annotation construct in
the GOOSDM schema allows the instances of CSG and ESG
to be mixed with the text content. The presence of this
construct along with the other defined constructs and
relationships, the proposed GOOSSDM is also capable to
model XML document at conceptual level.
VII. IMPLEMENTATION OF GOOSSDM USING GME
The Generic Modeling Environment (GME) provides meta-
modeling capabilities and where a domain model can be
configured and adapted from meta-level specifications
(representing the Conceptual modeling) that describe the
domain concept. It is common for a model in the GME to
contain several numbers of different modeling elements with
hierarchies that can be in many levels deep. The GME supports
the concept of a viewpoint as a first-class modeling construct,
which describes a partitioning that selects a subset of
conceptual modeling components as being visible.
Moreover, GME support the programmatic access of the
metadata of GME models. Most usual techniques for such
programmatic access is to write GME interpreter for some
metamodel. The interpreter will be able to interpret any domain
model based on that predefined metamodel. GME interpreters
are not standalone programs, they are components (usually
Dynamic Link Libraries) that are loaded and executed by GME
upon a user's request. Most GME components are built for the
Builder Object Network (BON), an inbuilt framework in GME
and provide a network of C++ objects. Each of these represents
an object in the GME model database. C++ methods provide
convenient read/write access to the objects' properties,
attributes, and relations described in GME metamodel.
In the context of GOOSSDM, the lower layers can be
conceptualized using levels in GME. The semi-structured data
Vol. 2, No. 10, 2011

120| P a g e

Figure11. Meta-Level Specifications of GOOSSDM model using GME

Figure 12. GOOSSDM Schema of Patient and Doctor Example using GME
definitions for any given GME model can be configured using
meta-level specifications of GOOSSDM. The interpreter will
generate the equivalent XML Schema Definitions for any given
GME model configured using meta-level specifications of
GOOSSDM to represent the semi-structured data at logical
level.
The meta-level specifications of GOOSSDM using GME
have been shown in Figure 11. The GOOSSDM schema
specification of Patient and Doctor example (Figure 9 and
Figure 10) using GOOSSFM meta-level specifications has been
shown in Figure 12. The BON based interpreter for
GOOSSDM can run from the GME interface to interpret any
GOOSSDM schema and to generate the equivalent XSD Code.
Vol. 2, No. 10, 2011

121| P a g e
VIII. CONCLUSION
In this paper, a model has been introduced for the
conceptual level design of semi-structured data using graph
based semantics. This is a comprehensive object oriented
conceptual model and the entire semi-structure database can be
viewed as a Graph (V, E) in layered organization. The graph
based semantics in GOOSSDM model extracts the positive
features of both Object and Relational data models and also it
maintains the referential integrity inherently. Further the
layered organization of the model facilitates to view the semi-
structured data schema from different level of abstraction.
The proposed GOOSSDM contains detailed set of
semantically enriched constructs and relationships those are
necessary to specify the facets of semi-structured database
system at conceptual level. Moreover, a set of rules also have
been proposed to systematically transform any GOOSSDM
schema to its equivalent XSD structure. The expressive powers
of the set of transformation rules have been illustrated with
suitable examples and case study. Moreover, the proposed
model also facilitates the designer to provide alternative design
of same schema by changing the ordering scheme, which in
result can be transformed in different XSDs with different
nesting patterns. It provides better understandability to the
users and high flexibility to the designers for creation and / or
modification of semi-structured data as well as XML document
at conceptual level. The proposed approach is also independent
from any implementation issues.
It is also important to note that with the concept of
Annotation construct the proposed GOOSSDM facilitate the
document centric design of semi-structured data at conceptual
level. Also the proposed model supports irregular,
heterogeneous, hierarchical and no-hierarchical structure in
data. Moreover, the set of proposed rules are capable to
transform systematically the GOOSSDM schema into
hierarchical XSD schema. Due to these features, the proposed
approach is also capable to design XML document at
conceptual level.
The proposed approach also has been automated through
the GME based meta-model configuration of GOOSSDM. The
meta-level specification of GOOSSDM along with interpreter
can be used as a CASE tool for the model by the semi-
structured database designer. The tools facilitates the automatic
generation of XML Schema Definitions from the conceptual
level graphical model, using the set of proposed rule set.
Future studies will concentrate on developing a graphical
query language for the proposed approach.
REFERENCES
[1] S. Abiteboul, P. Buneman, and D. Suciu, Data on the Web: From
Relations to Semistructured Data and XML, Morgan Kaufman, 1999.
[2] Jason McHugh, Serge Abiteboul, Roy Goldman, Dallas Quass, Jennifer
Widom "Lore: a database management system for semistructured data",
Vol. 26 , Issue 3, PP: 54 - 66, 1997
[3] A. Badia, "Conceptual modeling for semistructured data", In Proc. of
the Third International Conference on Web Information Systems
Engineering, PP: 170 177, 2002.
[4] M. Mani. EReX: A Conceptual Model for XML, In Proc. of the
Second International XML Database Symposium, PP 128-142, 2004.
[5] G. Psaila, ERX: A Conceptual Model for XML Documents, In Proc. of
the ACM Symposium on Applied Computing, PP: 898-903, 2000.
[6] A. Sengupta, S. Mohan, R. Doshi, XER - Extensible Entity Relationship
Modeling, In Proc. of the XML 2003 Conference, PP: 140-154, 2003.
[7] Martin Necasky, "XSEM: a conceptual model for XML", In Proc. of 4
th

ACM International Asia-Pacific conference on Conceptual Modeling,
Vol. 67, PP: 37 - 48, 2007.
[8] B. F. Lsio, A. C. Salgado, L. R. Galvo, "Conceptual modeling of
XML schemas", In Proc.of the 5
th
ACM International Workshop on Web
Information and Data Management, PP: 102 105, 2003.
[9] H. X. Liu, Y. S. Lu, Qing Yang, "XML conceptual modeling with
XUML", In Proc. of the 28
th
International Conference On Software
Engineering, PP: 973 976, 2006.
[10] C. Combi, B. Oliboni, "Conceptual modeling of XML data", In Proc. of
the ACM Symposium On Applied Computing, PP: 467 473, 2006.
[11] X. Wu, T. W. Ling, M. L. Lee, G. Dobbie, "Designing semistructured
databases using ORA-SS model", In Proc. of the 2
nd
International
Conference on Web Information Systems Engineering, Vol. 1, PP: 171
180, 2001.
[12] R. Conrad, D. Scheffner, J. C. Freytag, "XML conceptual modeling using
UML", In Proc. of the 19
th
International Conference On Conceptual
Modeling, PP: 558-574, 2000.
[13] M. Necasky, Conceptual Modeling for XML: A Survey, Tech. Report
No. 2006-3, Dep. of Software Engineering, Faculty of Mathematics and
Physics, Charles University, Prague, 2006.
[14] . Ldeczi, A. Bakay, M. Maroti, P. Volgyesi, G. Nordstrom, J.
Sprinkle, G. Karsai, Composing Domain-Specific Design
Environments, IEEE Computer, pp. 44-51, November 2001.
[15] A. Narayanan, G. Karsai, Specifying the correctness properties of
model transformations, Proc. of 3
rd
Int. workshop on Graph and model
transformations (Int. Conf. on Software Engineering), PP 45-52, 2008.
[16] Abhijit Sanyal, Anirban Sarkar, Sankhayan Choudhury, Automating
Web Data Model: Conceptual Design to Logical Representation, 19
th

Intl. Conf. on Software Engineering and Data Engineering (SEDE 2010),
PP 94 99, 2010.
[17] Anirban Sarkar, Sesa Singha Roy, Graph Semantic Based Conceptual
Model of Semi-structured Data: An Object Oriented Approach, 11
th

International Conference on Software Engineering Research and Practice
(SERP 11), Vol. 1, PP 24 30, USA, July 18 21, 2011.

AUTHORS PROFILE
Anirban Sarkar is presently a faculty member in the
Department of Computer Applications, National Institute of
Technology, Durgapur, India. He received his PhD degree
from National Institute of Technology, Durgapur, India in
2010. His areas of research interests are Database Systems
and Software Engineering. His total numbers of
publications in various international platforms are about 25.

IJACSA Volume2No10

Uploaded by

Copyright:

Available Formats

IJACSA Volume2No10

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

IJACSA Volume2No10

Uploaded by

Copyright:

Available Formats

(IJACSA) International Journal of Advanced Computer Science and Applications,

Vol. 2, No. 10, 2011

[support=1% and confidence=50%]

You might also like