Joint Cooperative Clustering and Power Control for Energy-Efficient Cell-Free XL-MIMO with Multi-Agent Reinforcement Learning

In this paper, we investigate the amalgamation of cell-free (CF) andextremely large-scale multiple-input multiple-output (XL-MIMO) technologies,referred to as a CF XL-MIMO, as a promising advancement for enabling futuremobile networks. To address the computational complexity and communicationpower consumption associated with conventional centralized optimization, wefocus on user-centric dynamic networks in which each user is served by anadaptive subset of access points (AP) rather than all of them. We begin ourresearch by analyzing a joint resource allocation problem for energy-efficientCF XL-MIMO systems, encompassing cooperative clustering and power controldesign, where all clusters are adaptively adjustable. Then, we propose aninnovative double-layer multi-agent reinforcement learning (MARL)-based scheme,which offers an effective strategy to tackle the challenges of high-dimensionalsignal processing. In the section of numerical results, we compare variousalgorithms with different network architectures. These comparisons reveal thatthe proposed MARL-based cooperative architecture can effectively strike abalance between system performance and communication overhead, therebyimproving energy efficiency performance. It is important to note thatincreasing the number of user equipments participating in information sharingcan effectively enhance SE performance, which also leads to an increase inpower consumption, resulting in a non-trivial trade-off between the number ofparticipants and EE performance.