System Information Feedback Assisted Reconfigurable Intelligent Surface Configuration: Leveraging Deep Reinforcement Learning

Reconfigurable intelligent surface (RIS) is a pivotal technology with considerable potential for the upcoming sixth-generation (6G) wireless communication systems, due to its ability to intelligently modify the characteristics of the radio environment. However, the current RIS configuration schemes, relying on perfect channel state information (CSI), are challenging to implement in practical settings. This paper proposes a deep reinforcement learning (DRL)-based framework for RIS configuration. The proposed framework employs a straightforward system information feedback mechanism to eliminate the need for perfect CSI. Initially, we model the interaction between the RIS and the wireless communication system as a Markov Decision Process (MDP). Subsequently, a RIS control scheme based on the twin delayed deep deterministic policy gradient (TD3) method is proposed. Unlike existing works, the proposed framework effectively adapts to the dynamic changes of the wireless channels. Finally, the simulation outcomes demonstrate that the proposed framework offers substantial improvements over existing benchmarks.