Security of Wireless Sensor Networks in the Era of Artificial Intelligence and IoT: A Systematic Literature Review
TL;DR
The review reveals several gaps in the research, such as the lack of explainability of models, use of benchmark datasets, susceptibility to adversarial and model-poisoning attacks, blockchain scalability issues, and the absence of standardized, secure system architectures that can offer reliable, privacy-preserving, and energy-efficient protection.
Abstract
The Wireless Sensor Networks (WSN) and the Internet of Things (IoT) have revolutionized various application areas such as smart cities, health, industrial automation, environment, agriculture, and intelligent transportation systems. Despite the successful widespread use of WSNs-IoT, they still have several security issues including resource constrained sensor nodes, decentralized design, and the combination of heterogeneous communication protocols and insecure wireless communication channels. Most traditional security solutions including cryptographic methods, intrusion detection systems are rule-based, which is not enough to protect against the advanced, evolving and zero-day attacks. Therefore, the paradigm of artificial intelligence (AI) has become an exciting approach to creating intelligent, adaptive and autonomous cyber security solutions. This paper is a systematic literature review of the security solutions based on artificial intelligence (AI) applied to WSNs (WSNs) in the context of IoT. Structured review methodology is followed in the study, which critically analyzes recent machine learning, deep learning, reinforcement learning, federated learning, blockchain and edge intelligence advancements in the field of intrusion detection, anomaly detection, threat prediction, authentication, privacy preservation, and secure communication. These approaches are compared on the basis of their accuracy to detect the target, computational complexity, energy efficiency, scalability, and privacy and feasibility for deployment in resource constrained environments. Moreover, it classifies the already known security threats, examines layer-wise defense measures and analyzes upcoming hybrid AI frameworks, which combine several intelligent technologies. The review reveals several gaps in the research, such as the lack of explainability of models, use of benchmark datasets, susceptibility to adversarial and model-poisoning attacks, blockchain scalability issues, and the absence of standardized, secure system architectures that can offer reliable, privacy-preserving, and energy-efficient protection. The paper then proposes future research directions that highlight the need of combining Explainable Artificial Intelligence (XAI), Federated Learning, Blockchain, and Edge AI for the construction of strong and adaptive cybersecurity frameworks. This review is a comprehensive reference for researchers and practitioners who are interested in designing secure, intelligent and sustainable WSN-IoT systems for next-generation cyber-physical ecosystems.