Movable Antenna-Enabled Secure Transmission for Active RIS-Aided ISAC Systems

The integrated sensing and communication (ISAC) technology empowered by movable antennas (MAs) has increased the possibility of wireless communication development. In this paper, we investigate a discrete MA-enabled active reconfigurable intelligent surface (RIS)-aided ISAC system. In this system, the position of discrete MAs can be dynamically adjusted to enhance channel conditions, thereby improving the communication security of legitimate users against potential eavesdropping risks from target. In particular, we formulate an optimization problem for the max-min secrecy rate (SR) by jointly optimizing beamforming matrix, phase shift matrix and MA discrete positions. An efficient alternating optimization (AO) based on penalty- successive convex approximation (SCA) and penalty-dual-decomposition (PDD) approach is proposed to obtain a high-quality initial solution for beamforming matrix and phase shift in the non-convex problem. Next, we propose a dueling double deep Q-Network (D3QN)-twin delayed and deep deterministic policy gradient (TD3) method to cope with the hybrid action space issue of the non-convex problem with highly coupled variables. Numerical results demonstrate that the introduction of MAs can effectively improve SR performance compared to fixed-position antenna (FPA) structure in RIS-aided ISAC systems. In addition, the proposed AO-based D3QN-TD3 method validate the suitability and superiority compared to the D3QN-TD3 method without AO algorithm.