Energy-Aware Resource Management for Federated Learning in Multi-Access Edge Computing Systems

In Federated Learning (FL), a global statistical model is developed by encouraging mobile users to perform the training on their local data and aggregating output parameters in an iterative manner. However, due limited energy computation capability at devices, performance of always stake meet objective minimization. this regard, Multi-access Edge Computing (MEC)-enabled FL addresses tradeoff between consumption devices allowing offload portion dataset edge server for training. Since has high capability, time insignificant. offloading from significant impact total consumed complete single round process. Thus, resource management MEC-enabled challenging, where reduce while saving devices. article, we formulate energy-aware which loss are jointly minimized, considering limitation addition, recast formulated problem as Generalized Nash Equilibrium Problem (GNEP) capture coupling constraints radio offloading. To that end, analyze computing allocation loss, time, consumption. Finally, present convergence analysis proposed solution, evaluate its against traditional approach. Simulation results demonstrate efficacy our solution