User Scheduling Algorithm Based on Parallel Monte Carlo Policy Gradient Sampling in Downlink Massive MIMO Communication Systems

As a pivotal technology for 5G and future mobile communication systems, massive multiple-input multiple-output (MIMO) has demonstrated the potential to substantially enhance spectral efficiency. However, the rapid proliferation of antennas and user equipment has rendered downlink user scheduling a critical challenge in massive MIMO systems. To address this issue, we propose a novel user scheduling algorithm designed to optimize system throughput while maintaining inter-user fairness. The proposed method formulates the user scheduling problem as a binary optimization problem and leverages a policy distribution-based sampling technique to determine the optimal scheduling solution. Furthermore, a dynamic adjustment mechanism for user fairness coefficients is incorporated to ensure equitable resource allocation. The numerical results show that the proposed algorithm improves the system sum rates by 59.76%, 116.48%, and 12.23% compared to the traditional proportional fair (PF) and round-robin (RR) algorithms, as well as a scheduling algorithm based on instantaneous channel rates (ICRS), respectively. Furthermore, Jain’s fairness index of the proposed algorithm reaches 98. 52%, 100. 11% and 98. 86% of those achieved by the PF, RR and ICRS algorithms, respectively.