Deep Learning and Optimization Approaches in Malicious Node Detection with Cross Attention Vision Transformers and Blockchain-Based Distributed Data Storage in Wireless Sensor Networks: A Review

Abstract

Wireless Sensor Networks (WSNs) have become integral to modern cyber-physical systems, enabling applications such as environmental monitoring, healthcare, and smart infrastructure. However, their distributed and resource-constrained nature makes them highly vulnerable to malicious node attacks, including black hole, Sybil, and denial-of-service attacks. Traditional security mechanisms fail to address these threats effectively due to scalability and adaptability limitations. This paper presents a comprehensive review of deep learning and optimization approaches for malicious node detection in WSNs, with a focus on Cross Attention Vision Transformers (CAVT) and blockchain-based distributed data storage. Deep learning models such as Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), and Graph Neural Networks (GNNs) have significantly improved detection accuracy by learning complex traffic patterns. Transformer-based models enhance this capability by capturing long-range dependencies and multi-modal relationships through attention mechanisms. Additionally, blockchain technology provides decentralized, tamper-proof storage and secure trust management among nodes. Optimization techniques further improve energy efficiency and network performance. This review analyzes recent studies, compares methodologies, identifies research gaps, and highlights future directions for building scalable and secure WSN architectures.