Volume 2, Issue 10, October 2012

ISSN: 2277 128X

International Journal of Advanced Research in

Fault Tolerance in Wireless Sensor Networks

Lambodar Jena

Aarti Pradhan

Dept. of CSE, GEC, BBSR

Dept. of CSE, GEC, BBSR

Dept. of CSE, GEC,BBSR

India

Abstract- Reliability is an important issue in a network and when the network is energy constrained as in a wireless
network then it becomes a high priority to ensure that this network is highly reliable. But this reliability is severely
affected by errors and faults that occur due to multiple reasons such as Hardware malfunctioning, bugs in software,
environmental hazards etc. Thus a sensor network should be fault resistant so that it can deal with these erroneous
conditions effectively. This paper tries to cover issues related to fault detection and recovery mechanisms in wireless
sensor networks in both theoretical and application oriented research.
Keywords- Fault Tolerant, Wireless sensor network, Multi- hop Routing, Replication, Aggregation
I.
INTRODUCTION
Advances in embedded systems technology have made it possible to build wireless sensor nodes, which are small devices
with limited memory, processing power, and energy resources [1]. Due to the low cost associated to these devices, it is
possible to conceive the deployment of large scale wireless sensor networks (WSN) with potentially thousands of nodes [7].
In critical environment a high degree of dependability is required. In order to be considered dependable, WSNs must offer
characteristics such as: reliability, availability and maintainability. Availability to a large extend depends on fault tolerance to
keep the system working as expected. Availability on the service level means that the service delivered by a WSN (or part of
it) is not affected by failures and faults in underlying components such as single nodes or node subsystems. In WSNs, the
failure of such components is almost unavoidable. Most detection and recovery techniques therefore aim at reducing MTTR
(the amount of time required for detecting and recovering from a failure) as much as possible. In these WSN deployments, it
is common to havea node providing functionality to its neighbors. Multi hop routing is a simple example of such a
service,where nodes forward messages on behalf of each other.Often, nodes assume dedicated roles such as cluster head,
which implies the responsibility for certain tasks. For example, a cluster head could aggregate sensor data beforeit is
forwarded to a base station, thereby saving energy.Nodes with stronger hardware capabilities can performoperations for other
nodes that would either have to spenda significant amount of energy or would not be capableof performing these
operations.These services, however, may fail due to various reasons, including radio interference, de-synchronization ,battery
exhaustion, or dislocation. Such failures arecaused by software and hardware faults, environmentalconditions, malicious
behavior, or bad timing of a legitimateaction. In general, the consequence of such an eventis that a node becomes unreachable
or violates certainconditions that are essential for providing a service, forexample by moving to a different location, the node
canno further provide sensor data about its former location.In some cases, a failure caused by a simple softwarebug can be
propagated to become a massive failureof the sensor network. This results in application trialsfailing completely and is not
acceptable in safety criticalapplications. Hence, our aim is to clarify the requirementsfor maintaining high level availability in
WSNs, and toinvestigate the tools and mechanisms utilized in WSNresearch and engineering for fault detection and recovery.
In this paper, we have made a detail analysis of various types of faults in a sensor network and have focused on
enhancements ofservice availability in WSNs through the use of fault tolerance techniques.We present a survey of
approachesto fault detection and recovery techniques in WSNs. Weprovide taxonomy of faults and classify the
investigatedapproaches according to their ability to detect and toleratefaults.
II.
FAILURES IN WIRELESS SENSOR NETWORKS
To comprehend fault tolerance mechanisms, it is important to point out the difference between faults, errors, and failures.
Various definitions of these terms have been used [3], [40]. This paper refers to the definition given in [40]:
A fault is any kind of defect that leads to an error.
An error corresponds to an incorrect (undefined) system state. Such a state may lead to a failure.
A failure is the (observable) manifestation of an error, which occurs when theSystem deviates from its specification and
cannot deliver its intended functionality. Figure 2 illustrates the difference between fault, error, and failure. A sensor service
running on node A is expected to periodically send the measurements of its sensors to an aggregation service running on node
B. However, node A suffers an impact causing a loose connection with one of its sensors. Since the code implementing node

2012, IJARCSSE All Rights Reserved

Page | 146

Sushruta et al., International Journal of Advanced Research in Computer Science and Software Engineering 2 (10),
October- 2012, pp. 146-153
As service is not designed to detect and overcome such situations, an erroneous state is reached when the sensor service tries
to acquire data from the sensor. Due to this state, the service does not send sensor data to the aggregation service within the
specified time interval. This results in a crash or omission failure of node A observed by node B.In the scenario explained
above, the fault is the loose connection of the sensor.

Node A
Sensing
task

Node B
Aggregation
task

Sensor
Fig. 1 A loosely connected sensor causing Failure

The error is the state of the service after trying to read the sensor data and the failure occurs when the application does not
send the sensor data within the specified time interval. To provide resilience in faulty situations two main actions must be
performed:
Fault detection. To provide any countermeasures, the first step a system must perform is to detect that a specific
functionality is or will be faulty.
Fault Recovery. After the system has detected a fault, the next step is to prevent or recover from it. Themain technique to
achieve this goal is to replicate the components of the system that are vital for its correct operation.
III.
Source of Faults in Real WSN Applications
Ultimately, if the application in the back-end which presents the WSN data to the users suffers a fault due to some software
bug or hardware failure the entire system is considered faulty. In this paper, however, we will concentrate on faults that can
happen in the sensor nodes up to the sink.
1) Node Faults: Nodes have several hardware and software components that can produce malfunctions. For example, the
enclosure can suffer impacts and expose the hardware of the sensor node to the extreme conditions of the environment. In
[27][41][39] due to stress from the environment and inadequate enclosures, the sensor nodes were exposed to direct contact
with water causing short circuits. The report of a large-scale deployment in a potatoes field [20] indicated that the antennas
from the nodes were quite fragile and would become loose when inserting the node into the packaging. When the battery of a
node reaches a certain stage, sensor readings may become incorrect. This has been observed in [43] where many outlier
readings were generated in the network caused by imminent battery failure. As demonstrated in Figure 3, hardware failures
will generally lead to software failure. A Data Acquisitionapplication will not perform properly if the underlying sensors are
providing incorrect readings. Nevertheless,some hardware failures do not affect all the services in asensor node. In the
example discussed, although the nodecannot be used to provide correct sensor readings it still can be used to route packages
in the sensor network.Software bugs are a common source of errors in WSNs. In [44], the researchers reported that a software
bug caused the longest continuous network outage taking the system offline for three days until the nodes could be
reprogrammed manually. Organizing a network in clusters is an approach used in many applications, for example to extend
the lifetime of the network [15]. A small number of nodes are selected to become clusterheads. They are responsible for
coordinating the nodes in their clusters, for instance bycollecting data from them and forwarding it to the base station. In case
that a clusterhead fails, no messages of its cluster will be forwarded to the base station any longer. The clusterhead can also
intentionally or due to software bugs forward incorrect information. Depending on the application case, the impact of such a
failure can vary from quality degradation of measurements to alarm messages not being delivered to a back-end system.
While forwarding messages, nodes can aggregate data from multiple other nodes in order to reduce the amount of data sent to
the base station. One common simple approach is to calculate the average of correlated measured values such as temperature,
humidity and pressure, sending only one message to the back-end. If a node generates incorrect data, the data aggregation
results can suffer deviations from the real value. Also, if a node responsible for generating the aggregated data is subject to a
value failure, the base station will receive incorrect information of an entire region of the network.
2) Network Faults: Routing is one of the fundamental building blocks in a WSN. It is essential for collecting sensor data,
distributing software and configuration updates, and for coordination among nodes. Additionally, there may be applicationspecific routing protocols required, for example for tracking and following moving objects. Faults on the routing layer can
lead to dropped or misguided messages, or unacceptable delays. In WSNs, communication links between nodes are highly
2012, IJARCSSE All Rights Reserved

Page | 147

Sushruta et al., International Journal of Advanced Research in Computer Science and Software Engineering 2 (10),
October- 2012, pp. 146-153
volatile. WSNs not always yield the same delivery rate of messages in field trials as in lab trials. For instance, in [38] a
delivery rate of only 58% of the messages was observed, in [36] the instability of the links between nodeslead to constant
changes in the routing paths. In several scenarios of sensor networks nodes have a certain degree of mobility. In a glacial
expedition [27] the experiment assumed a one hop network. The connection of the nodes to the sink was calibrated during
deployment with a reliable link connection. Nevertheless, after some time the nodes moved to a different location where the
node was unreachable resulting in complete loss of data from this node. Radio interference can also cause the link between
nodes to become faulty. For instance, in agricultural fields the placement of the nodes must be carefully planned to take into
consideration that when plants start growing the link range is considerably reduced, as discussed in [42]. Another source of
link failure is the collision of messages. In [39] researchers observed a potential for collision of messages of nodes in close
proximity due to a phase change and overlap. In other situations, however, nodes may have perfect link connections but the
messages are not delivered to their destination due to path errors. A software bug in the routing layer can generate circular
paths or simply deliver messages to the incorrect destination.
3) Sink Faults: On a higher level of the network a device (sink) that collects all the data generated in the network and
propagates it to the back-end system is also subject to faults of its components. When this device fails, unless fault tolerant
measurements are present, a massive failure of the network happens given that the data from the sensor nodes cannot be
accessed. The sink can be deployed in areas where no permanent power supply is present, in such applications batteries
together with solar cells are commonly applied [27][20][43] to provide the amount of energy necessary. In the glacial
expedition reported in [27], this traditional technique has proven to be inefficient. Although this worked perfectly for other
expeditions, in this glacial environment the sink suffered a power failure due to snow covering the solar cells for several
days. Network infrastructure is usually also not available in the area where the sink is deployed, therefore alternative
solutions such as a satellite connection are used, which cause fluctuations in the back-end network can interface. In [24]
researchers indicated that during periods of severe thunderstorm activity the satellite connection becomes unavailable.
Finally, the software that stores the data collected from the network, processes it and sends it to the back-end system, is
subject to bugs that when present can lead to loss of data within the period where the fault occurred
IV.
Failure Classification
As discussed in Section IV-A, several faults could lead to failures in wireless sensor networks: a node could be moved to a
different region, causing a link failure; nodes can suffer power failure and stop responding to requests, or they can start
sending arbitrary values either intentionally (after a security breach) or due to a malfunction. Here, we classify the failures
that a WSN is susceptible to, which are: crash, omission, timing, value and arbitrary. These failures are the observable
manifestation of underlying faults presented on Section IV-A.
1) Crash or omission: A failure by omission is determined by a service sporadically not responding to requests. For instance,
this could be caused by radio interference that leads to occasional message loss. A crash failure occurs when the service at
some point stops responding to any request. An omission degree f can be defined which imposes a limit to the amount of
omission failures a node might have before being classified as crashed.
2) Timing: Services might fail due to a timeout in processing a request or by providing data too early. Such timing failures
occur when a node responds to a request with the correct value, but the response is received out of the time interval specified
by the application. Timing failures will only occur when the application specifies timing constraints.
3) Value: A service is considered having failed due to an incorrect value when the service sends a timely response, however
with lack in accuracy of the delivered value. For instance, a service performing aggregation of data generated by other nodes
could forward a result value to the base station that does not correctly reflect the input data. Such situations could be caused
by malfunctioning software, hardware, corrupt messages, or even malicious nodes generating incorrect data.
4) Arbitrary: Arbitrary failures include all the types of failures that cannot be classified in previously described categories. In
[19], Lamport has introduced the Byzantine Generals Problem in the context of distributed systems. Recent work shows how
to deal with this problem in the domain of wireless sensor networks [16]. Byzantine failures describe a type of arbitrary
failures that are in general caused by a malicious service that not only behaves erroneously, but also fails to behave
consistently when interacting with other services and applications. In sensor networks, an aggregation service could start
sending both incorrect and correct values to the sink, ora node routing messages could not forward a messagedespite sending
an acknowledge back to the sender.
V.
FAULT DETECTION TECHNIQUES
The goal of fault detection is to verify that the services being provided are functioning properly, and in some cases to predict
if they will continue to function properly in the near future. The simplest way to perform such a task is through visual
observation and manual removal of incorrect values. This technique has obvious drawbacks: human interaction leads to
errors, it has a high cost and it is not efficient. Hence, we investigated automatic fault detection techniques for WSN. We
classified the techniques we investigated according to the parties involved in the process. Through selfdiagnosis the node
itself can identify faults in its components. With group detection, several nodes monitor the behavior of another node.
Finally, in hierarchical detection the fault detection is performed using a detection tree where a hierarchy is defined for the

2012, IJARCSSE All Rights Reserved

Page | 148

Sushruta et al., International Journal of Advanced Research in Computer Science and Software Engineering 2 (10),
October- 2012, pp. 146-153
identification of failed nodes. Often in a hierarchical detection the detection is shifted to a more powerful device such as the
sink.
A. Self-Diagnosis
In many cases, nodes can identify possible failures by performing self-diagnosis. In [14], the authors propose an approach
where a node would perform a self-diagnosis based on the measurements of accelerometers to determine if the node suffers
from an impact that could lead to hardware malfunctions.
Using a similar approach, nodes could detect when they are being moved to a different location. Another
approach would be to keep track of the identities of the nodes in the neighborhood. A considerable change in the
neighborhood could indicate that either the node itself or some of its previous neighbors have been moved. Faults caused by
battery exhaustion can be predicted when the hardware allows the measurement of the current battery voltage [2][29]. By
analyzing the battery discharge curve and the current discharge rate, an algorithm can determine an estimation of the time to
death of the battery. Nodes can also identify that their current connection to surrounding nodes is unreliable by probing the
link connection therefore identifying that it is isolated.
B. Group Detection
The detection of services failing due to incorrectly generated values is only possible if a reference valueis available. In [18]
and [6], detection mechanisms are proposed to identify faulty sensor nodes. Both algorithms are based on the idea that
sensors from the same region should have similar values unless a node is at the boundary of the event-region. The algorithm
start by taking measurements of all neighbors of a node and uses the results to calculate the probability of the node being
faulty. Another approach proposed in the literature is to let consumer nodes observe whether the service provider is in fact
performing the operations that it is supposed to. In [26], a misbehavior detection algorithm to aide the routing layer is
proposed. The misbehavior detection mechanism is based on the idea of monitoring thecommunication of the service
provider to verify whether messages are forwarded correctly. Focusing on providing a fault-tolerant approach for clusters in
WSNs, in [12] it is proposed to support the dynamic recovery of failed gateways (high-energy devices that act as
clusterheads). The proposed protocol assumes that a gateway has failed only when no other gateways can communicate with
it. The fault detection mechanism is based on constant status updates being exchanged between gateways and further use of a
consensus algorithm.
C. Hierarchical Detection
The definition of a detection tree enables a scalable fault detection algorithm in WSN. Memento [33] proposes the usage of
the network topology to forward the fault detection results of child nodes to the parent nodes and up to the sink. Each node
forwards the status of the child nodes that it is monitoring to its parent node. The parent performs an aggregation (bitwise
OR) operation on the results of the child nodes together with its on results and forwards it to the next level. The approach
proposed by Memento scales well with the network size, however it consume resources of the network. Shifting the fault
detection task to a more powerful device is an alternative that can help to increase the lifetime of the WSN. In [37], the
authors propose an algorithm that puts the burden of detecting and tracing failed nodes to the base station. At first the nodes
learn the network topology and send their portion of the topologyinformation to the base station. With this information the
base station learns the complete network topology which is used to send route updates as soon as it detects that nodes become
silent. This approach is not applicable to event-driven WSN because in such a network sensors only send messageswhen
there is an event that should be reported, for instance when the temperature goes above a certain limit. In [35], the authors
focus on providing a solution in this context. The proposed mechanism uses a hierarchical network topology where cluster
heads monitor ordinary nodes, and the base station monitors the cluster heads. To perform the monitoring, the base station
and the clusterheads constantly ping those nodes that still have battery power left and that are under their direct supervision.
If a node does not respond, it is marked as failed. Sympathy [30] is a debuging tool that also utilizes the hierarchical detection
approach. This tool instruments the WSN with monitoring software on the sensor nodes that generates metrics data that is
forwarded to a centralized sink location for analysis.With this information Sympathy is able to detect crash, timeout and
omission failures and identify the fault that generated the failure.
VI.
FAULT RECOVERY TECHNIQUES
Fault recovery techniques enable systems to continue operating according to their specifications even if faults of a certain
type are present. As discussed in section IV, there are many potential sources for faults in WSNs. Fault tolerance techniques
have been proposed in various contexts that increase the reliability of the functionality of sensor nodes in their specific
domain. We attempt at giving an overview of this scattered work. The most common of these techniques is the replication of
components. Although redundancy has several advantages in terms of high reliability and availability, it also increases the
costs of a deployment. As an alternative, according to the specification of the project, the quality required from the WSN can
be downgraded to an acceptable level. In this paper we classify the recovery techniques forWSN into two major approaches:
Active and Passive replication. Active replication means that all requests are processed by all replicas, while with passive
replication, a request is processed by a single instance and only when this instance fails, another instance takes over.
A. Active replication in WSN
Active replication in wireless sensor networks is naturally applied in scenarios where all, or many, nodes provide the same
functionality. One example is a service that periodically provides sensor data. Nodes that run this service activate their

2012, IJARCSSE All Rights Reserved

Page | 149

Sushruta et al., International Journal of Advanced Research in Computer Science and Software Engineering 2 (10),
October- 2012, pp. 146-153
sensors and forward their readings to an aggregation service or to a base station. When some nodes fail to provide that
information, the recipient still gets the results from other nodes, which is often sufficient. Fault recovery in the presence of
active replicas is relatively straightforward. Nevertheless, for a consistent survey we present some of these approaches here:
1) Multipath routing: Usually, it is desirable to avoid that a single failing node causes the partitioning of asensor network.
Thus, a network should be k-connected, which allows k 1 nodes to fail while the network would still be connected [22].
Multipath routing [10] can be used to actively replicate routing paths. In [4], Bredin et al. proposes an algorithm that
calculates the minimum amount of additional nodes and their positions to guarantee k-connectivity between nodes.
2) Sensor value aggregation: Sensor value fusion [28][5] is a reasearch area that seeks to provides highlevel information
derived from a number of low-level sensor inputs. There, the inherent redundancy of sensor nodes can be used to provide
fault-tolerant data aggregation.This is achieved through a tradeoff between the precision (the length) of the resulting sensor
reading interval and the number of faulty sensors. This ensures that despite of node failures, the resulting reading interval will
contain the correct sensor reading of a region.
3) Ignore values from faulty nodes: A simple but efficient solution to not propagate a failure of one specific node to the
entire network is to ignore the data that it is generating, as applied in [14]. The major challenge in this case is the
identification of the malfunctioning nodes.
B. Passive replication in WSN
When passive replication is applied, the primary replicareceives all requests and processes them. In order tomaintain
consistency between replicas, the state of theprimary replica and the request information are transferredto the backup
replicas. Given the constraints of WSNs, applicationsshould be designed to have only little or no stateat all, which minimizes
the overhead for transferring stateinformation between nodes or eliminates it altogether.The process of recovering from a
fault when usingpassive replication is illustrated in Figure 4 and consistsof three main steps: fault detection (discussed in
SectionV, primary selection and service distribution.
1) Node selection: After it has been established thata certain functionality is not available any longer due toa failure in the
primary replica, a new service providermust be selected. After this selection phase, one or severalnodes become service
providers. Several approaches tohow the selection is performed have been proposed. Wedifferentiate them according to who
makes the decisionon which party should become a service provider.
a) Self election: In LEACH [15], nodes periodically execute a probabilistic algorithm to establish whether they should serve
as clusterhead to their neighbors. In this probabilistic rotation system, nodes keepchanging their role in the network. When a
clusterhead node fails, it will take only one rotation period until another node starts providing the functionality of the failed
or absent node.

Fig. 2 Failure recovery in Passive replication mode

b) Group election: In [12], a reallocation of nodes that were part of a cluster that suffered a clusterhead failure is proposed.
The clusterhead, called gateway, is considered to be a resourceful node. The solution presented considers that all the
gateways in the network maintain a list of the nodes that are currently in their cluster and another backup list of nodes that
could become part of their cluster. When a gateway fails, the nodes from its cluster are reallocated to the other gateways that
have the nodes in their backup lists. If more than one gateway has a specific node in its backup list the node is assigned to the
clusterhead that has the smallest communication cost.
c) Hierarchical election: In a hierarchical election, a coordinator selects the new primary node. This applies to the rebuilding
of routing paths [37] as well as the selection of a new clusterhead [13]. The former describes an algorithm to select the node
that is closest to the base station. The latter approach applies fuzzy logic in the base station to select which node will become
a clusterhead. This algorithm makes use of a fuzzy descriptor, the node concentration, energy level in each node and its
centrality with respect to the entire cluster.
2) Service Distribution: During this phase, nodes elected to become service providers must activate the

2012, IJARCSSE All Rights Reserved

Page | 150

Sushruta et al., International Journal of Advanced Research in Computer Science and Software Engineering 2 (10),
October- 2012, pp. 146-153
service. In some cases the service is already available on the nodes and a simple configuration change to inform the node that
this service should be activated is required. However in some cases, for instance when nodes do not have enough memory to
store the code of all potential services, it is necessary to inject code into the node through some technique. There are different
techniques that can be used for service distribution: completely reprogramming the node, sending entire blocks of executable
code, or sending small pieces of code such as scripts.
Pre-Copy: Pre-copying as described in [9], consists in making the code of all services available on all nodes before
deployment. This allows nodes to change their behavior according to the role that they are assigned to.
b) Code distribution: Several approaches have been proposed for disseminating code throughout the network. Mate [21] is
an example for a bytecode interpreter for TinyOS where code is broken into capsules of 24 instructions. These capsules can
be distributed through the network and installed on nodes, which start to execute the new code. Agilla [8] is a Mate-based
mobile agent middleware for programming wireless sensor networks. These mobile agents can be programmed to move
through the network or replicate themselves to other nodes according to changes in the environment.
c) Remote Execution: On the one hand code migrationis an approach that reduces the amount of memoryrequired in the
entire network since not all nodes needto have the application pre-installed, on the other handit consumes energy on the
nodes exchanging the codeand is susceptible to link failures, which could causelong delays until the code update is
completed. RemoteExecution [34], [32] is an alternative approach where lowpowerdevices transfer tasks to more powerful
deviceswithout transferring the entire application code. Instead,only the required state information is transmitted. Suchan
approach is especially suited for heterogeneous sensornetworks with at least some resourceful nodes.
VII.
CONCLUSION
In this paper we provided a thorough investigation of faults that occured in real WSN deployments. This concise
investigation provides a valuable knowledge input for future application to prevent the same kind of issues from happening.
By focusing only on the faults, the lessons learned from the different deployments can be used by any application even if the
investigated trial had a different research focus. We proposed taxonomy to classify faults and failures that occur in WSN. We
also studied the problem of fault detection and recovery, surveying the different techniques currently applied in WSN
research. A classification of the available fault tolerance techniques for wireless sensor networks has been proposed
considering the various mechanisms adopted by the solutions. To our knowledge this is the first work that provides a concise
survey and classification in this area. Through the classification proposed it is possible to compare the different solutions
identifying the strong and weak points of each of them. This allows for a correct selection of the techniques that are more
suitable to specific applications.

REFERENCES
[1] I. F. Akyildiz, W. Su, Y. Sankarasubramaniam, and E. Cayirci. ASurvey on Sensor Networks. IEEE Communications
Magazine, 40:102114, August 2002.
[2] L. Benini, G. Castelli, A. Macii, E. Macii, M. Poncino, and R. Scarsi.A Discrete-Time Battery Model for High-Level
Power Estimation. In Proceeding of the Design, Automation and Test in Europe Conference and Exhibition 2000, pages 35
39, 2000.
[3] A. Birolini. Quality and Reliability of Technical Systems:theory, practice, management. Springer, 1997.
[4] J. L. Bredin, E. D. Demaine, M. Hajiaghayi, and D. Rus. Deploying Sensor Networks with Guaranteed Capacity and Fault
Tolerance. In Proceedings of the 6th ACM international symposium on Mobile ad hoc networking and computing, pages
309319, 2005.
[5] D. Desovski, Y. Liu, and B. Cukic. Linear randomized voting algorithm for fault tolerant sensor fusion and the
corresponding reliability model. In IEEE International Symposium on Systems Engineering, pages 153162, October 2005.
[6] M. Ding, D. Chen, K. Xing, and X. Cheng.Localized fault-tolerant event boundary detection in sensor networks.In
INFOCOM, 2005.
[7] D. Estrin, R. Govindan, J. Heidemann, and S. Kumar. Next century challenges: scalable coordination in sensor networks.
In MobiCom99: Proceedings of the 5th annual ACM/IEEE international conference on Mobile computing and networking,
pages 263270,
New York, NY, USA, 1999. ACM Press.
[8] C.-L. Fok, G.-C.Roman, and C. Lu. Mobile Agent Middleware for Sensor Networks: an Application Case Study. In
Proceedings of the 4th International Symposium on Information Processing in Sensor Networks (IPSN), 2005.
[9] C. Frank and K. Romer. Algorithms for Generic Role Assignment in Wireless Sensor Networks. In Proceedings of the
3rd international conference on Embedded networked sensor systems, pages 230242, 2005.
[10] D. Ganesan, R. Govindan, S. Shenker, and D. Estrin. Highly- Resilient, Energy-Efficient Multipath Routing in Wireless
Sensor Networks.Mobile Computing and Communications Review, 1(2), 1997.
[11] R. Guerraoui and A. Schiper.Fault-Tolerance by Replication in Distributed Systems. In Proceedings of the 1996 AdaEurope International Conference on Reliable Software Technologies, pages 3857, 1996.
2012, IJARCSSE All Rights Reserved

Page | 151

Sushruta et al., International Journal of Advanced Research in Computer Science and Software Engineering 2 (10),
October- 2012, pp. 146-153
[12] G. Gupta and M. Younis.Fault-Tolerant Clustering of Wireless Sensor Networks. Wireless Communications and
Networking, 3:15791584, 2003.
[13] I. Gupta, D. Riordan, and S. Sampalli. Cluster-Head Election Using Fuzzy Logic for Wireless Sensor Networks. In
Proceedings of the 3rd Annual Communication Networks and Services Research Conference, pages 255260, 2005.
[14] S. Harte and A. Rahman. Fault Tolerance in Sensor Networks Using Self-Diagnosing Sensor Nodes. In The IEE
International Workshop on Intelligent Enviroment, pages 712, June 2005.
[15] W. R. Heinzelman, A. Chandrakasan, and H. Balakrishnan. Energy-Efficient Communication Protocol for Wireless
Microsensor Networks. In Proceedings of the 33rd Hawaii International Conference on System Sciences, volume 8, page
8020, 2000.
[16] C.-Y. Koo. Broadcast in Radio Networks Tolerating Byzantine Adversarial Behavior. In Proceedings of the twenty-third
annual ACM symposium on Principles of distributed computing, pages 275282, 2004.
[17] F. Koushanfar, M. Potkonjak, and A. Sangiovanni-Vincentell. Fault Tolerance Techniques for Wireless Ad hoc Sensor
Networks. In Proceedings of IEEE Sensors, volume 2, pages 14911496, 2002.
[18] B. Krishnamachari and S. Iyengar. Distributed Bayesian Algorithms for Fault-Tolerant Event Region Detection in
Wireless Sensor Networks. IEEE Transactions on Computers, 53:241250, March 2004.
[19] L. Lamport, R. Shostak, and M. Pease. The Byzantine Generals Problem. ACM Transactions on Programming
Languages and Systems, 4:382401, 1982.
[20] K. Langendoen, A. Baggio, and O. Visser. Murphy loves potatoes: experiences from a pilot sensor network deployment
in precision agriculture. In IPDPS 20th International Parallel and Distributed
Processing Symposium, 2006.
[21] P. Levis and D. Culler.Mate: A Tiny Virtual Machine for Sensor Networks. In ASPLOS-X: Proceedings of the 10th
international conference on Architectural support for programming languages and operating systems, pages 8595, New
York, NY, USA, 2002. ACM Press.
[22] N. Li and J. C. Hou. FLSS: A Fault-Tolerant Topology Control Algorithm for Wireless Networks. In Proceedings of the
10th Annual International Conference on Mobile Computing and Networking, pages 275286, 2004.
[23] T. Liu and M. Martonosi. Impala: A Middleware System for Managing Autonomic, Parallel Sensor Systems. In PPoPP
03: Proceedings of the ninth ACM SIGPLAN symposium on Principles and practice of parallel programming, pages 107
118, New York, NY, USA, 2003. ACM Press.
[24] A. Mainwaring, D. Culler, J. Polastre, R. Szewczyk, and J. Anderson. Wireless sensor networks for habitat monitoring.
In WSNA 02: Proceedings of the 1st ACM international workshop on Wireless sensor networks and applications, pages 88
97, New York, NY, USA, 2002. ACM Press.
[25] D. Marculescu, N. H. Zamora, P. Stanley-Marbell, and R. Marculescu. Fault-Tolerant Techniques for Ambient
Intelligent Distributed Systems. In Proceedings of the 2003 IEEE/ACM international conference on Computer-aided design,
page 348, 2003.
[26] S. Marti, T. J. Giuli, K. Lai, and M. Baker. Mitigating Routing Misbehavior in Mobile Ad hoc Networks. In Proceedings
of the 6th annual international conference on Mobile computing and networking, pages 255265, 2000.
[27] K. Martinez, P. Padhy, A. Riddoch, H. Ong, and J. Hart. Glacial environment monitoring using sensor networks. In
REALWSN 05, 2005.
[28] K. Marzullo. Tolerating failures of continuous-valued sensors. ACM Transactions on Computer Systems, 8(4):284304,
1990.
[29] D. Rakhmatov and S. B. Vrudhula. Time-to-Failure Estimation for Batteries in Portable Electronic Systems. In
Proceedings of the 2001 international symposium on Low power electronics and design, pages 8891, 2001.
[30] N. Ramanathan, K. Chang, R. Kapur, L. Girod, E. Kohler, and D. Estrin. Sympathy: a debugging system for sensor
networks. In IEEE International Conference on Local Computer Networks, 2004.
[31] M. Ringwald, K. Romer, and A. Vitaletti.Snif: Sensor network inspection framework. Technical Report 535, ETH
Zurich, Institute for Pervasive Computing, October 2006.
[32] P. Rong and M. Pedram. Extending the lifetime of a network of battery-powered mobile devices by remote processing:
A markovian decision-based approach. In DAC 03: Proceedings of the 40 th conference on Design automation, pages 906
911, New York, NY, USA, 2003.ACM Press.
[33] S. Rost and H. Balakrishnan. Memento: A health monitoring system for wireless sensor networks. In SECON, 2006.
[34] A. Rudenko, P. Reiher, G. J. Popek, and G. H. Kuenning. The Remote Processing Framework for Portable Computer
Power Saving. In SAC 99: Proceedings of the 1999 ACM symposium on Applied computing, pages 365372, New York,
NY, USA, 1999. ACM Press.
[35] L. B. Ruiz, I. G. Siqueira, L. B. e Oliveira, H. C. Wong, J. M. S. Nogueira, and A. A. F. Loureiro. Fault Management in
Eventdriven Wireless Sensor Networks. In Proceedings of the 7th ACM
international symposium on Modeling, analysis and simulation of wireless and mobile systems, pages 149156, June 2004.
[36] T. Schmid, H. Dubois-Ferri`ere, and M. Vetterli. Sensorscope: Experiences with a wireless building monitoring sensor
network. In REALWSN 05, 2005.
2012, IJARCSSE All Rights Reserved

Page | 152

Sushruta et al., International Journal of Advanced Research in Computer Science and Software Engineering 2 (10),
October- 2012, pp. 146-153
[37] J. Staddon, D. Balfanz, and G. Durfee. Efficient Tracing of Failed nodes in Sensor Networks. In Proceedings of the 1st
ACM international workshop on Wireless sensor networks and applications, pages 122130, 2002.
[38] R. Szewczyk, A. Mainwaring, J. Polastre, J. Anderson, and D. Culler. An analysis of a large scale habitat monitoring
application. In SenSys 04: Proceedings of the 2nd international conference on Embedded networked sensor systems, pages
214 226, New York, NY, USA, 2004.ACM Press.
[39] R. Szewczyk, J. Polastre, A. M. Mainwaring, and D. E. Culler. Lessons from a sensor network expedition. In EWSN,
pages 307 322, 2004.
[40] A. S. Tanenbaum and M. van Steen. Distributed Systems: Principles and Paradigms. Prentice Hall, 2002.
[41] J. Tateson, C. Roadknight, A. Gonzalez, T. Khan, S. Fitz, I. Henning, N. Boyd, and C. Vincent. Real world issues in
deploying a wireless sensor network for oceanography. In RealWSN 05, 2005.
[42] J. Thelen, D. Goense, and K. Langendoen. Radio wave propagation in potato fields. In First Workshop on Wireless
Network Measurement, 2005.
[43] G. Tolle, J. Polastre, R. Szewczyk, D. Culler, N. Turner, K. Tu, S. Burgess, T. Dawson, P. Buonadonna, D. Gay, and W.
Hong. A macroscope in the redwoods. In SenSys 05: Proceedings of the 3rd international conference on Embedded
networked sensor systems, pages 5163, New York, NY, USA, 2005. ACM Press.

2012, IJARCSSE All Rights Reserved

Page | 153