RIS-Assisted MIMO Secure Communication Schemes Based on Fixed and Dynamic Subarray Partition

The subarray partition of reconfigurable intelligent surface (RIS) can reduce the complexity and hardware cost of the RIS architecture while achieving satisfactory performance, which has great practical significance. However, existing RIS subarray partition methods do not consider the scenario with eavesdropper and are not suitable for such scenario. Therefore, this paper discusses the joint optimization design of subarray partition and reflection coefficients to maximize the system secrecy rate in the multi-antenna eavesdropper scenario. Since this problem is a nonconvex combinatorial optimization problem, we solve it by alternating optimization, and propose RIS-assisted MIMO secure communication schemes based on fixed and dynamic subarray partition respectively. In fixed subarray partition scheme, the closed expressions of the optimal transmit covariance matrix and the optimal reflection coefficients are obtained through theoretical analysis, where the reflection coefficients are optimized by coordinate descent method. The fixed subarray partition scheme is not limited by the subarray structure and the number of subarray units. In dynamic subarray partition scheme, we transform the subarray partition optimization problem into a task assignment problem, and propose an improved discrete particle swarm optimization (DPSO) algorithm to solve it. Simulations and analyses show that compared with the without partition scheme, two proposed schemes can achieve a satisfactory secrecy rate while reducing system complexity and hardware cost. In addition, the dynamic subarray partition scheme has better secrecy rate performance and better stability than fixed subarray partition scheme, but its implementation is more complex.