A Novel Energy-Efficiency Optimization Framework for Wireless-Powered RIS-Assisted Communication Systems

In this paper, we study a wireless-powered Reconfigurable Intelligent Surface (RIS)-assisted communication system, where multiple users upload data to a base station (BS) with the help of an RIS and the users and the RIS are powered by the energy harvested from the radio frequency signals emitted by the BS. Unlike conventional passive RISs adopted in the literature, we employ a hybrid RIS in the system, of which each reflecting element is enabled to operate in energy-harvesting (EH) mode, passive mode or active mode. Our goal is to maximize the energy efficiency of the system, by jointly optimizing the operating modes of the reflecting elements at the RIS and resource allocation in the system. The formulated optimization problem is hard to solve directly. Therefore, we propose to solve the problem with a deep reinforcement learning approach. The simulation results show that our proposed approach can significantly improve the energy efficiency of the system.