基于深度强化学习的RIS辅助无蜂窝大规模MIMO系统电磁干扰能量收集方法

Electromagnetic interference (EMI) poses a significant negative impact on reconfigurable intelligent surface (RIS)-assisted communications. In contrast, in this correspondence, we harness EMI for energy harvesting (EH) in RIS-assisted cell-free (CF) massive multiple-input multiple-output (mMIMO) systems. The proposed scheme achieves energy-efficient RISs by employing a dynamic time-division strategy to split the time into transmission and EH phases. It promotes environmentally sustainable communication by efficiently harnessing the power of EMI and access points (APs). Specifically, we jointly optimize the precoding of APs, the phase shifts of RISs, and transmission time scheduling to maximize EH while satisfying the required quality of service (QoS) for users. Due to the non-convex and non-linear nature of the problem, arising from the intricate coupling of optimization variables, a novel deep reinforcement learning (DRL) algorithm based on the twin-delayed deep deterministic policy gradient (TD3) is proposed for robust system performance. Simulation results demonstrate the proposed scheme effectively reduces the negative effects of EMI and optimally utilizes the power of EMI, especially in scenarios with severe EMI.