Multi-metric trust-based secure routing with intelligent anomaly detection for robust IoT network communication

Abstract

The rapid expansion of the internet of things (IoT) has introduced significant challenges in ensuring secure, reliable, and energy-efficient communication across heterogeneous networks. This paper proposes a multi-metric trust based secure routing (MMTSR) framework that combines comprehensive trust evaluation with intelligent anomaly detection to mitigate routing attacks and enhance overall network performance. The trust evaluation model aggregates multiple behavioral and performance metrics, including packet forwarding ratio (PFR), residual energy (RE), packet loss rate (PLR), communication delay (CD), and historical interaction reliability (HIR), to compute a composite trust score for each node. Simultaneously, a machine learning-based anomaly detection module identifies malicious nodes based on traffic behavior, routing updates, and packet transmission patterns. Simulation results demonstrate that MMTSR significantly outperforms conventional routing protocols, achieving a packet delivery ratio of 97.8%, average throughput of 470 kbps, and end-to-end delay of 46 ms under a 10% malicious node scenario, while reducing routing overhead and energy consumption to 150 packets and 0.95 J per node, respectively. The anomaly detection module achieves a detection accuracy of 97.6% with a false positive rate of only 2.5%, effectively identifying blackhole, selective forwarding, Sybil, and packet dropping attacks. Comparative analysis with recent studies further confirms the framework’s superiority in terms of reliability, security, and efficiency.