Comprehensive Survey of Deep Learning-Based Intrusion Detection for Securing Routing in IoT Wireless Sensor Networks

Abstract- The proliferation of Internet of Things (IoT) Wireless Sensor Networks (WSNs) incritical
sectors demands robust security solutions to counter complex routing attacks such as sinkhole, blackhole, and selective forwarding. Conventional detection methods often fall short inresource-constrained, dynamic IoT WSN environments. Deep Learning (DL) techniquesincluding Convolutional Neural Networks (CNN), Recurrent Neural Networks (RNN), andAutoencoders have shown remarkable ability to autonomously detect and classifyroutingsecurity threats with high accuracy. This comprehensive survey systematically reviews recent DL-based Intrusion DetectionSystems (IDS) designed to secure routing in IoT WSNs. It examines key DL architectures, evaluates their performance using benchmark datasets and metrics, and discusses challengesincluding real-time deployment, energy efficiency, and model interpretability. Finally, thepaper outlines future research directions toward developing lightweight, adaptive, anddistributed DL-IDS specifically tailored for IoT WSNs.

Keywords- Deep Learning, Intrusion Detection System, IoT Wireless Sensor Networks, Routing Security, Routing Attacks, CNN, RNN, Autoencoders, Lightweight Models