Optimizing Energy Efficiency in Cell-Free Massive MIMO Systems: A Distributed DNN-Driven Method

This letter explores the downlink power allocation challenge for cell-free massive multiple-input multiple-output (MIMO) systems. Focusing on energy efficiency (EE) optimization, we propose a distributed power allocation framework driven by deep neural networks (DNN). This method employs an unsupervised learning strategy and relies solely on local large-scale fading channel statistics to operate efficiently. Moreover, the proposed method demonstrates adaptability to scenarios with varying numbers of access points (APs). Numerical results highlight the remarkable advantages of the proposed approach relative to various benchmark schemes.