Deep Reinforcement Learning-Based Intelligent Reflecting Surface for Secure Wireless Communications

In this paper, we study an intelligent reflecting surface (IRS)-aided wireless secure communication system for physical layer security, where IRS is deployed to adjust its elements guarantee of multiple legitimate users in the presence eavesdroppers. Aiming improve secrecy rate, a design problem jointly optimizing base station (BS)'s beamforming and IRS's formulated given different quality service (QoS) requirements time-varying channel condition. As highly dynamic complex, it challenging address non-convex optimization problem, novel deep reinforcement learning (DRL)-based approach firstly proposed achieve optimal policy against eavesdroppers environments. Furthermore, post-decision state (PDS) prioritized experience replay (PER) schemes are utilized enhance efficiency performance. Specifically, PDS capable tracing environment characteristics accordingly. Simulation results demonstrate that PDS-PER learning-based can significantly rate QoS satisfaction probability IRS-aided systems.