An Evolutionary Game Theory and Reinforcement Learning-Based Security Protocol for Intermittently Connected Wireless Networks

Intermittently Connected Wireless Networks (ICWNs) are characterized by dynamic node mobility and the absence of persistent end-to-end paths, making them highly susceptible to security threats. This paper proposes a novel secure routing protocol, called the Evolutionary Game Theoretic model with Reinforcement Learning (EGT-RL), designed to provide adaptive and resilient protection against blackhole attacks in such networks. EGT-RL integrates Q-learning for dynamic threat assessment with evolutionary game theory to model and influence node behavior over time. Simulation results, based on both synthetic and real-world mobility traces, show that EGT-RL significantly outperforms three benchmark protocols in delivery ratio, packet drops, end-to-end latency, and communication overhead.