Energy Efficiency Optimization for User-Heterogeneous RIS-Aided CF-mMIMO System Based on DDPG

Cell-free massive multiple-input multiple-output (CF-mMIMO) provides service through multiple distributed access points (APs), improving the performance of traditional base station systems. However, in complex environments, such as densely populated urban areas, the presence of signal barriers and dead zones can reduce communication quality. To address the issue of communication quality degradation caused by signal obstructions, this paper constructs a reconfigurable intelligent surface assisted cell-free massive multiple-input multiple-output (RIS-CF-mMIMO) communication system that supports heterogeneous users. Different from existing research, this paper considers modeling user heterogeneity using different numbers of antennas in RIS-CF-mMIMO systems, aiming to maximize the energy efficiency (EE) of the RIS-CF-mMIMO system for heterogeneous users. To achieve this, we propose a scheme named heterogeneous deep deterministic policy gradient optimized beamforming and phase shift matrix (Het-DDPG-BF-PSM). This scheme utilizes the deep deterministic policy gradient (DDPG) algorithm to jointly optimize the receiving and transmitting beamforming matrices as well as the RIS phase shift matrix. The simulation results show that compared with existing methods, the proposed Het-DDPG-BF-PSM scheme significantly improves EE of CF-mMIMO communication systems and demonstrates its adaptability in complex and heterogeneous environments.