Multi-RIS-aided covert communication for ISAC systems: performance analysis and DRL-based algorithm

Abstract

Multi-Reconfigurable Intelligent Surface (RIS)-aided wireless networks has demonstrated the potential advantage in improving capacity and degree of freedom. In this paper, we investigate a novel multi-RIS-aided covert communication for Integrated Sensing And Communication (ISAC) system, where the signal of the Base Station (BS) transmit through multiple RISs to estimate the target Direction-Of-Arrival (DoA) and send message to Bob covertly against the Willie’s surveillance. We established performance boundary optimization problems centered on communication and sensing, where solutions were obtained by employing a combination of Successive Convex Approximation (SCA) and Semi-Definite Relaxation (SDR) schemes, as well as a low complexity method, respectively. Moreover, we jointly design the transmit beamforming and the multi-RISs’ phase shifts, as well as receiving beamforming to maximize the sensing-centric Energy Efficiency (EE). To deal with the intractable optimization problem, the Enhanced Distributional Soft Actor-Critic (E-DSAC) based Deep Reinforcement Learning (DRL) algorithm was employed to obtain the optimal transmission strategy to prevent eavesdropping. Numerical results show the performance upper limit of multi-RIS-aided ISAC systems and demonstrate significant gains of E-DSAC scheme on sensing-centric EE over other DRL based algorithms.