Secure Robust Resource Allocation Algorithm for RIS-Assisted MIMO Systems

‍ ‍The rapid development of communication services has made wireless communication systems more demanding in terms of spectral efficiency and system capacity. A Reconfigurable Intelligent Surface （RIS） can reconfigure the wireless transmission environment by adjusting the reflection coefficients of the reflection units， thus effectively improving the system performance and spectrum efficiency. However， the openness of the wireless channel makes it impossible to guarantee the security of the transmitted data. Physical Layer Security （PLS） technology fully leverages the physical characteristics of the wireless channel to encrypt the transmission information， which is a crucial means to solve the problem of wireless communications being vulnerable to eavesdropping attacks from the physical layer， but its security performance is affected by the Channel State Information （CSI） of the user. Aiming at the high spectral efficiency demand of wireless communication systems and the large loss of system security performance caused by the imperfect CSI of users， an RIS-assisted Multiple Input Multiple Output （MIMO） system model with the presence of multiple eavesdropping users is constructed， and a security robust resource allocation algorithm is proposed. First， a resource allocation model that jointly optimizes the Base Station （BS） antenna power allocation and RIS passive beamforming to maximize the secrecy rate of legitimate users is developed for the imperfect CSI of eavesdropping users under the BS transmit power and RIS phase shift constraints. Subsequently， an algorithm for solving the nonconvex problem with infinitely many nonconvex constraints for the above multivariate coupling is proposed. For problems that are difficult to solve with multivariate coupling， the original problem is transformed into two subproblems of BS emission covariance matrix optimization and RIS phase-shift optimization using alternate optimization method， and the problem is completed by transforming it into a convex optimization problem using the Charnes-Cooper transform， S-Procedure method， and the convex difference algorithm based on the penalty function， respectively. Finally， the feasibility and effectiveness of the algorithm are proven by convergence and complexity analysis. Simulation results show that the proposed algorithm has a better legitimate user secrecy rate and robustness in the presence of multiple eavesdropping users and a bounded uncertainty of the eavesdropping channel.