Energy Efficiency Optimization for Federated Learning in Cell-free Massive MIMO Systems

In recent years, federated learning (FL), a promising machine learning technique, has gained increasing attention. Unlike traditional approaches that centralize data for training a global model, FL enables model training to occur on user devices (UEs), prioritizing data privacy and security. A key challenge in implementing FL over wireless communication lies in optimizing the energy efficiency of user devices, especially considering their limited power resources. In this paper, we address the problem of minimizing total energy consumption during the training phase of FL in a cell-free massive multiple-input multiple-output system. The optimization problem is formulated with the goal of minimizing the overall energy consumption of all UEs during the FL training process. To address the nonconvex and intricate problem, we propose an iterative algorithm utilizing the inner approximation method. This approach transforms the problem into a series of successive convex problems, yielding at least one locally optimal solution. Extensive experiments show the effectiveness of the proposed algorithm in addressing the given problem. The results demonstrate the superior performance of the proposed approach compared with benchmark algorithms.