STAR-RIS-Assisted Sum Rate Maximization for Underlaid D2D Communications

作者：Huy T. Nguyen, D. Nguyen, T. Nguyen · 发表于：Conference on Research, Innovation and Vision for the Future in Computing & Communication Technologies · 年份：2025 · DOI：10.1109/rivf68649.2025.11365048

This article investigates the use of a simultaneous transmitting and reflecting reconfigurable intelligent surface (STAR-RIS) to improve coverage and spectral efficiency in a device-to-device (D2D) underlaid cellular network. We analyze a scenario where a multi-antenna base station (BS) simultaneously transmits messages to multiple primary users, while multiple D2D pairs reuse the same spectrum. This work focuses on the joint optimization of several key elements such as BS beamforming, the STAR-RIS phase-shift coefficients, D2D transmit power, and the selection of active STAR-RIS elements. The goal is to maximize the achievable sum-rate for all downlink users while effectively managing interference and satisfying constraints on the BS power budget and each user's minimum rate threshold. Due to the high-dimensional and non-convex nature of this problem, traditional optimization methods face limitations in dynamic and large-scale environments. Therefore, we propose a deep reinforcement learning (DRL) framework using Proximal Policy Optimization (PPO) to tackle the system uncertainty. Simulation results show our proposed PPO-based DRL algorithm converges well and outperforms several baselines, i.e., approximated 27% to Advantage Actor-Critic (A2C) and 45% to Soft Actor-Critic (SAC) models.