Long-Term CSI-based Design for RIS-Aided Multiuser MISO Systems Exploiting Deep Reinforcement Learning

TL;DR

This paper proposes a long-term CSI-based transmission design for RIS-aided multiuser MISO systems, utilizing deep reinforcement learning to optimize beamforming and phase shifts, resulting in higher net throughput compared to traditional methods.

Contribution

It introduces a novel DDPG-based algorithm for long-term CSI-based transmission design in RIS systems, reducing channel estimation overhead and improving throughput.

Findings

Achieves higher net throughput than conventional schemes.

Effectively reduces channel estimation overhead.

Demonstrates the effectiveness of deep reinforcement learning in RIS design.

Abstract

In this paper, we study the transmission design for reconfigurable intelligent surface (RIS)-aided multiuser communication networks. Different from most of the existing contributions, we consider long-term CSI-based transmission design, where both the beamforming vectors at the base station (BS) and the phase shifts at the RIS are designed based on long-term CSI, which can significantly reduce the channel estimation overhead. Due to the lack of explicit ergodic data rate expression, we propose a novel deep deterministic policy gradient (DDPG) based algorithm to solve the optimization problem, which was trained by using the channel vectors generated in an offline manner. Simulation results demonstrate that the achievable net throughput is higher than that achieved by the conventional instantaneous-CSI based scheme when taking the channel estimation overhead into account.