Intelligent Reconfigurable Surface Assisted Secure Communication System Based on DDPG Algorithm

Intelligent Reconfigurable Surface (IRS), as an important technology in future communication networks, has received extensive attention. This paper investigates IRS assisted secure communication system, where multiple eavesdroppers (eves) exist. Under the maximum transmission power of the base station and the constraints of the IRS unit mode, the base station transmission beamforming vector and the IRS phase shift vector are jointly optimized. To solve this nonconvex problem, we proposed a deep deterministic policy gradient (DDPG) algorithm grounded in deep reinforcement learning (DRL). The simulation results show that the communication system with IRS has a high confidentiality rate, reflecting the advantages of IRS in the communication system. Moreover, our proposed method based on DRL has an acceptable speed of convergence.