AI-Driven Secure Multipath Routing With Energy-Efficient Lifetime Optimization For Iot Networks

Abstract

The fast proliferation of Internet of Things (IoT) deployments has brought serious issues related to secure data routing, node energy exhaustion, network lifetime decay, and network resource-heavy wireless settings. Traditional single-path and multi-path routing algorithms do not consider optimal path discovery, security implementation, and energy conservation. It proposes EMRA-IoT, an AI-driven secure multipath routing framework that integrates Deep Q-Network reinforcement learning for optimal multipath discovery. The framework also incorporates a Trust-Integrated Cryptographic Security Enforcement Layer, combining ECC-160 and HMAC-SHA256 authentication, an LSTM-driven predictive traffic load balancing module, and an Adaptive Node Duty Cycle Control mechanism for maximizing residual energy. It is mathematically modeled as a Markov Decision Process formulation, composite reward functions, and energy dissipation equations. The simulation outcomes show a Packet Delivery Ratio of 96.7%, an end-to-end delay of 48.2 ms, a network lifetime of 497 seconds, and 48.9% less energy consumption than the current protocols. The proposed framework offers simultaneous routing optimality, lightweight security, and energy efficiency in next-generation IoT networks.