RL DDPG-Based Adaptive Access Point Selection for Cell-Free Massive MIMO Systems in 6G Networks

Abstract

This paper proposes a Deep Deterministic Policy Gradient (DDPG)-based adaptive access point (AP) selection framework for cell-free massive multiple-input multiple-output (CF-mMIMO) systems in beyond-5G and 6G networks. Unlike conventional full-cooperation or heuristic clustering methods, the proposed approach learns a continuous AP weighting policy that dynamically selects a subset of APs based on channel conditions and spatial distribution. A reward function jointly optimizing spectral efficiency, AP locality, and utilization efficiency is introduced. Simulation results demonstrate that the proposed framework achieves high spectral efficiency with significantly reduced AP activation, leading to improved energy efficiency and reduced fronthaul overhead. The learned policy exhibits spatially localized cooperation, confirming the effectiveness of reinforcement learning for scalable CF-mMIMO optimization.