AI-Based Secure Routing: Intrusion Detection for IoT Networks

Security threats in IoT networks, particularly routing attacks such as Sybil, sinkhole, blackhole, and wormhole attacks, compromise data integrity and network performance. To mitigate these threats, we propose AIRS (AI-Driven Intrusion-Resilient Secure Routing), an AI-powered secure routing algorithm that integrates machine learning-based anomaly detection with a trust-aware routing mechanism to enhance network resilience. AIRS was implemented in the Cooja simulator and evaluated against Secure-RPL and Trust-Based LEACH. Simulation results show that AIRS achieves 96.5% intrusion detection accuracy, improves packet delivery ratio (94%), and reduces energy consumption (1.3J per node), leading to a 40% increase in network lifetime compared to Secure-RPL. Additionally, AIRS minimizes false alarms, achieving a false positive rate of 3.5%, reducing unnecessary security overhead. The results demonstrate that AI-driven intrusion detection enhances routing security and efficiency in IoT networks. AIRS provides a scalable, energy-efficient, and resilient security solution for large-scale IoT deployments. Future work will focus on federated learning-based adaptive detection and real-world implementation to further validate AIRS in dynamic IoT environments.