Papers by Yıldıran Yılmaz
Neural Computing and Applications
Aims: To propose a deep learning algorithm for pothole detection and compare the performance of S... more Aims: To propose a deep learning algorithm for pothole detection and compare the performance of Sigmoid and Softmax activation functions in the creation of Convolutional Neural Network (CNN) algorithms. Methods: Three different datasets were used to justify the robustness of the CNN model in detecting dry and wet potholes. The CNN algorithms were created separately using the Sigmoid and Softmax activation functions. Results: The CNN algorithm using the Sigmoid function achieved higher accuracy scores than the CNN algorithm using the Softmax function. Specifically, the Sigmoid algorithm achieved accuracy scores of 91%, 96%, and 83% over datasets 1, 2, and 3, respectively, while the Softmax algorithm achieved scores of 81%, 96%, and 85% over the same datasets. Conclusion: The results of this study suggest that the CNN algorithm using the Sigmoid activation function is more robust and effective in detecting pothole images compared to the CNN algorithm using the Softmax activation function.
At present, the number of passengers preferring to use the airline is increasing with each passin... more At present, the number of passengers preferring to use the airline is increasing with each passing day. Thus, correctly analysing the airfare prices is essential to raise awareness of passengers. Some researchers have applied different kinds of Machine Learning (ML) algorithms to predict the airfare prices. However, to the best of our knowledge, penalized regression methods have not been used to analyse the airfare prices. Ridge, Lasso, and Elastic Net regressions are penalized regression methods. The dataset used in this study consists of 1814 one-way flights from Greece to Germany. The developed Ridge, Lasso, and Elastic Net methods were achieved to provide convincing results for airfare prices analysis based on Mean
IEEE Transactions on Emerging Topics in Computing, Oct 1, 2021
Counterfeited products are costing the global economy hundreds of billions of dollars annually. R... more Counterfeited products are costing the global economy hundreds of billions of dollars annually. Radio frequency identification(RFID) technology provides a promising solution for this problem, wherein each product is fitted with a secure tag, which is difficult to forge. However, RFID technology is faced with numerous security threats, for example, if the communication link between the reader and the tag is compromised, then it will be possible for a malicious adversary to obtain the private data stored on the device. Tag cloning attacks have also been demonstrated to be feasible, which severely undermines the capabilities of the RFID technology to protect against counterfeiting. One solution to this problem is the use of authentication protocol; however, existing schemes do not support mutual authentication and are still vulnerable to tag cloning attacks. In this paper, a new security mechanism is proposed, which consists of a lightweight three-flights mutual authentication protocol and an anti-counterfeit tag design. The proposed solution is based on combining the Rabin public-key encryption scheme with physically unclonable functions (PUF) technology. The security of the proposed protocol is systematically analysed and compared with existing schemes. The implementation cost of the proposed security primitives, assuming the 1024-bit public key, is 10139 GEs, which is suitable for low-cost RFID tags. Our results show that the proposed design is up-to 50% more area-efficient compared to systems based on Elliptic Curve Cryptography (ECC).
Authentication is a fundamental part of essential security operations and is a cornerstone for th... more Authentication is a fundamental part of essential security operations and is a cornerstone for the Internet of Things (IoT) security. In this work an energy-efficient and secure mutual authentication protocol is proposed for constrained IoT devices wherein a combination of RC5 (Rivest Cipher) and ECC (Elliptic Curve Cryptography) cryptosystems are used. The protocol is implemented, and its functionality is verified on Zolertia RE-mote IoT devices. It supports secure data transmission along with authentication. Unlike existing schemes, mutual authentication in the proposed protocol is achieved with only two flights between client and server. The security against most common attacks is analysed, furthermore energy consumption of our protocol is evaluated and compared with existing protocol e.g. DTLS handshake. Our protocol saves up to 57% energy compared to the DTLS handshake protocol per authentication cycle.
Journal of building engineering, Oct 1, 2023
Hitite journal of science and engineering, Jun 30, 2022
T he Distributed Denial-of-Service (DDOS) attack, which is a type of cyber-attack based on preven... more T he Distributed Denial-of-Service (DDOS) attack, which is a type of cyber-attack based on preventing a device or network resources from being accessed by temporarily or indefinitely disrupting the services of a host connected to the Internet, causes enormous economic losses [1]. For example, the DDOS attacker sends excessive requests to the target web server in IoT and prevents this website from working correctly by exceeding its capacity to process multiple requests. Producing a real-time attack detection system with a low cost in terms of computational burden remains one of the foremost challenges. Internet of Things (IoT) devices pose a greater risk than other computing devices in public networks because firmware updates and maintenance are not accomplished on most IoT devices after deployment [2].
The Internet of Things (IoT) devices, especially in wireless sensor networks (WSN) and radio freq... more The Internet of Things (IoT) devices, especially in wireless sensor networks (WSN) and radio frequency identification (RFID) systems, have limited resources. For example, battery-operated sensor devices forming the basic structure of the wireless sensor networks and the passive tags forming the basic structure of the RFID system can now be produced in smaller sizes and lower cost because of the development in electronics. Therefore, the wireless sensor devices are limited in memory, computation ability and energy resource and passive tags are limited in their chip area and computation ability. However, these devices make the task of establishing security protocols difficult as they dedicate most of the available resources to perform their basic tasks. On the other hand, these devices are based on pervasive technology which makes it easy for potential adversities and third parties to have access to them and carry well-established attacks. Security protocols that rely on storing keys or secrets in the devices' memory cannot provide physical security, as the key or secret data stored on the device is always open to physical tampering. Physical Unclonable Functions (PUF) technology is a potential candidate cryptographic primitive to incorporate physical security into the system. However, PUF-based protocols come with complications. Their security has not been established against communication traffic attacks. The machine learning and the Man in the Middle (MitM) are considered major security threats for PUF-based protocols. Consequently, there are two challenges in developing and implementing security protocols specific to resource-constrained devices; the first challenge is to provide security against common attacks, including physical attacks. The second challenge is to make the use of resources efficient considering the device-specific characteristics for the smooth implementation of the security protocol. This thesis proposes three new security protocols in consideration of the unique characteristics of IoT devices. The proposed and implemented security protocols that ensure privacy and mutual authentication in wireless sensor devices and RFID system are described in details, and their security is analysed systematically. The first protocol is named ASSURE and is based on PUF technology and a symmetric light cipher. It provides security for wireless sensor devices and ensures communication traffic privacy, physical security and energy-efficiency. The second security protocol is called ARMOR and is based on PUF and lightweight asymmetric encryption. It provides security for the RFID system and ensures communication traffic privacy and physical security. For the ARMOR protocol, it is proposed a PUF enabled area-efficient tag architecture. The third protocol termed TiGHTEN provides energy-efficiency and mutual authentication for wireless sensor devices and it consumes less energy compared to the first protocol. In the third protocol aimed at minimising communicational energy consumption, both symmetric (RC5) and asymmetric cryptosystems (ECC) are used to provide mutual authentication.
DergiPark (Istanbul University), Jun 30, 2021
H umanity has faced an outbreak (Covid-19) first time after the second world war. The origin of C... more H umanity has faced an outbreak (Covid-19) first time after the second world war. The origin of Covid-19 is Wuhan, China and it was announced initially in December 2019 [1]. By 6 January 2021, the spreading of this outbreak is quite fast around the world, and more than 87.2 million people have infected over 192 countries [2]. Besides, 1.88 million people have died due to Covid-19. To date, various timeseries forecasting models have played an important role in the maintenance of health public systems. Additionally, governments obtain information about the spread and consequences of infectious diseases through these models. In this sense, governments have taken some precautions to reduce the spreading speed of Covid-19 based on the results of these models, such as declare a curfew at regular intervals, quarantines etc. [3]. In the field of case forecasting, various studies have been proposed to predict Covid-19 cases. More detailed information about the proposed studies are available in the literature (Literature Review Section.).
IEEE Access, 2022
Public key algorithms are heavily used in many digital applications including key establishment s... more Public key algorithms are heavily used in many digital applications including key establishment schemes, secure messaging apps, and digital signature schemes in cryptocurrencies. Recent developments in the field of quantum computation have placed these algorithms at risk as they enable the implementation of more effective attacks to derive the secret key. Most notably Shor's algorithm exponentially speeds up solving the factoring, discrete logarithm (DLP), and elliptic-curve discrete logarithm (ECDLP) problems. To address this challenge, NIST has initiated a process to develop and standardize a new quantum-resistant publickey cryptographic algorithm. However, asymmetric encryption schemes are known to be computationally intensive, hence energy demanding. The proliferation of energy-constrained internet of things devices, combined with the need to adopt higher complexity quantum resilient cryptographic algorithms, makes it more challenging to continue to use public-key algorithms for all applications. One approach to address these challenges is to adopt symmetric key systems, which are known to be more energy-efficient and more resilient to quantum computers-based attacks. This work performs a comprehensive comparison of energy costs between asymmetric and symmetric key schemes. This comparison is performed using two methods. The first approach uses the energy cost of data usage (ECDU) metric to evaluate the global energy costs associated with internet data usage. It was found that the annual energy consumed by applications associated with public-key cryptography globally is sufficient to provide electricity for 1000 UK households for a year. The second method uses an experimental technique based on constructing a small-scale network of wireless embedded devices. This is subsequently used to compare two key establishment schemes, symmetric and asymmetric, which allows for comparing the computation and communication costs of each solution in a controlled environment, and more importantly estimating the energy consumed by each device participating in the protocol. Our results show that a 58% saving in global energy costs of public key-based applications can be achieved by adopting symmetric key systems. It was also found that a 20% reduction of the energy consumed by a wireless device during a key agreement protocol, can be achieved if symmetric key encryption is used. INDEX TERMS Symmetric-key encryption, public-key cryptography, key exchange protocols, digital signatures, energy.
Computer Science
Providing machine learning (ML) based security in heterogeneous IoT networks including resource-c... more Providing machine learning (ML) based security in heterogeneous IoT networks including resource-constrained devices is a challenge because of the fact that conventional ML algorithms require heavy computations. Therefore, in this paper, lightweight ProtoNN, CMSIS-NN, and Bonsai tree ML algorithms were evaluated by using performance metrics such as testing accuracy, precision, F1 score and recall to test their classification ability on the IPv6 network dataset generated on resource-scarce embedded devices. The Bonsai tree algorithm provided the best performance results in all metrics (98.8 in accuracy, 98.9% in F1 score, 99.2% in precision, and 98.8% in recall) compared to the ProtoNN, and CMSIS-NN algorithms.
Hittite Journal of Science and Engineering
Internet of Things that process tremendous confidential data have difficulty performing tradition... more Internet of Things that process tremendous confidential data have difficulty performing traditional security algorithms, thus their security is at risk. The security tasks to be added to these devices should be able to operate without disturbing the smooth operation of the system so that the availability of the system will not be impaired. While various attack detection systems can detect attacks with high accuracy rates, it is often impos-sible to integrate them into Internet of Things devices. Therefore, in this work, the new Distributed Denial-of-Service (DDoS) detection models using feature selection and learn-ing algorithms jointly are proposed to detect DDoS attacks, which are the most common type encountered by Internet of Things networks. Additionally, this study evaluates the memory consumption of single-based, bagging, and boosting algorithms on the client-side which has scarce resources. Not only the evaluation of memory consumption but also development of ensemble learni...
DergiPark (Istanbul University), Nov 30, 2021
88 yerine çevrimiçi olması nedeniyle öğrencilerin performansının düştüğünü ortaya koymaktadır. Bu... more 88 yerine çevrimiçi olması nedeniyle öğrencilerin performansının düştüğünü ortaya koymaktadır. Bu yüzden, bu çalışma, bilgisayar programlama modülünün öğretiminde çevrimiçi eğitime göre daha etkili ve verimli bir yöntem olan yüz yüze/geleneksel eğitimin önemini vurgulamıştır.
Dünya Multidisipliner Araştırmalar Dergisi, Nov 30, 2021
Uploads
Papers by Yıldıran Yılmaz