Physical Layer Security-Based Relay Selection for Wireless Cooperative Networks: A Reinforcement Learning Approach

This paper introduces a reinforcement learning-based physical layer secured relay selection (RS) scheme for the dual-hop non-regenerative relaying system. The physical layer security (PLS) techniques are either used to boost the performance of the upper layer security techniques or as an alternative to the classical security techniques. The PLS-based relay selection degrades the performance of the adversaries and hence enhances the secrecy of the system. The emerging and future wireless technologies are moving toward machine learning, and consequently, we proposed a Markov decision process-based Q-learning algorithm to train the agent for best relay selection in cooperative wireless networks. Q-learning is a less complex ML algorithm that introduces self-learning in each relay node and requires a small amount of information exchange between the nodes. The benefit of the proposed algorithm has been verified through simulation results.