Aerial Reconfigurable Intelligent Surface-Assisted Secrecy: A Learning Approach

In this letter, we propose to use an aerial reconfigurable intelligent surface (ARIS) to enhance the physical layer security and defend against an eavesdropper. We jointly consider the aerial deployment and passive beamforming of the ARIS to maximize the secrecy rate. In particular, we decompose the problem into two-layer problem, where the passive beamforming as inner problem is solved through semidefinite programming and relaxation and the aerial deployment as outer problem is tackled with deep Q-learning. Simulation results are provided to show the deployment and security performance, which demonstrate the superiority of our proposal as compared with the baselines in terms of secrecy rate.