Abstract
Reconfigurable Intelligent Surfaces (RIS) are emerging technology to enhance the 6G wireless communication systems by intelligently reconfiguring the propagation environment. This paper propose a Modified Prioritized Deep Deterministic Policy Gradient (MP-DDPG) algorithm for jointly optimizing beamforming at the Base Station (BS) and RIS phases in a Multiple Input Single Output (MISO) downlink communication system. The primary objective is to minimize transmitted powers while adhering to critical constraints such as maximum power limits and Quality of Service (QoS) requirements of the User's Equipment (UE). The simulation results show that the MP-DDPG algorithm offers a robust and adaptive framework for addressing the inherent non-convexity and dynamic nature of such an optimization problem. Key findings highlight the remarkable resilience to imperfect Channel State Information (CSI), and the crucial trade-offs between performance, computational complexity, and signaling overhead.