Secrecy Throughput Maximization for RIS-Assisted FBL Massive MIMO Systems

This letter investigates the physical security of reconfigurable intelligent surface (RIS) aided massive multi-input multi-output (MIMO) downlink eavesdropping system for multi-users with finite blocklength (FBL). Based on statistical channel state information (CSI), we formulate a joint optimization problem of the blocklength and phase shift matrix to maximize the user's secrecy throughput (ST), aiming to ensure the trade-off between security and throughput. However, due to the non-convex feature of challenging problem, we propose the momentum stochastic successive convex approximation (M-SSCA) algorithm. To accelerate convergence, the information from the previous optimization steps is utilized by the momentum parameter, which can aid in the faster discovery of the global optimum solution. Numerical results demonstrate the benefit and effectiveness of the M-SSCA phase optimization over the existing baseline schemes.