ENHANCING SECURITY IN WIRELESS SENSOR NETWORKS INTEGRATED TO INTERNET OF THINGS USING DIFFERENT SCHEMES

Abstract

This work focuses on enhancing security in wireless sensor networks (WSNs) integrated into the Internet of Things (IoT) by employing different schemes. WSNs are widely used in various fields, such as defence, transportation, healthcare, and environmental monitoring, where they collect and process data from the surrounding environment. However, due to the nature of WSNs and resource constraints, they are vulnerable to security threats. Attacks such as Sybil attacks, routing attacks, and various other forms of malicious activities can compromise the integrity and reliability of WSNs. The strategies and frameworks proposed in this thesis aim to overcome these security issues. The proposed schemes include robust intrusion detection systems, localization techniques based on range-free and range-based strategies, secure clustering, data aggregation, routing, and optimization techniques. These schemes utilize machine learning models, such as hybrid machine learning algorithms, to detect and classify attacks. The performance of the proposed models is evaluated using benchmark datasets, considering metrics such as training and testing time, precision, recall, F1-score, and accuracy. The proposed research aims to address security challenges and vulnerabilities present in IoT based WSNs by employing the design of advanced intrusion detection systems to detect and mitigate both internal and external attacks in IoT-based WSNs. These systems are designed to identify and respond to malicious activities that can compromise the integrity and functionality of the network. Ensemble Machine learning techniques are utilized for classifying and detecting DoS attacks using benchmark datasets. Integrating trust evaluation methodologies into the localization process helps to solve the security concerns that arise as a result of the procedure. DV-Hop, RSSI, and DE localization techniques assess the nodes' trustworthiness in localization, considering factors such as their behaviour, reliability, and communication history. By evaluating the trustworthiness of nodes, the localization process can mitigate the impact of malicious nodes and ensure the accuracy and integrity of localization results. viii The hybrid DV-Hop-RSSI-DE localization approach is coupled with the MLPANN machine learning algorithm and achieves better localization accuracy to detect malicious nodes. MLPANN is trained using labelled datasets to identify patterns indicative of malicious behaviour. The system can detect and classify malicious nodes based on their localization characteristics by leveraging machine learning. Furthermore, as WSNs are an integral part of the IoT, this thesis also explores security challenges specific to IoT-based WSNs. A hierarchical design incorporating blockchain-based cascaded encryption and trust evaluation is proposed to improve security and service delivery in IoT-WSNs. By combining raw data from devices and identifying risks, federated machine learning improves data security and transport. The proposed approach demonstrates improved performance and security in large-scale IoT-WSNs, leveraging heterogeneous wireless sensor networks to provide secure services. Secure data aggregation and clustering techniques are also proposed to detect and classify attacks in WSNs. These techniques optimize data aggregation to extend the network lifetime and outperform existing security and performance metrics approaches. These techniques minimize communication overhead and maximize resource utilization while protecting sensitive information using hybrid GA-PSO based on fuzzy rule intelligent routing protocol in WSNs. Lastly, trust management and routing mechanisms enable secure and efficient communication in distributed and hierarchical network topologies. These mechanisms ensure that data is routed through trustworthy nodes and establish reliable communication paths within the network. The effectiveness of the proposed schemes and frameworks is validated through simulations and comparisons with recent works. The results confirm the improved security and performance achieved by the proposed methods. By addressing security concerns through intrusion detection, secure localization, machine learning, secure clustering, and trust management, the security and reliability of IoT-based WSNs can be significantly enhanced. The scalability and applicability of the techniques in large-scale deployments are also addressed. This research contributes to developing secure and reliable WSNs integrated into the IoT and provides avenues for future work.