Federated Learning-Empowered Resource Allocation Optimisation for E-Health IoT System

This paper proposes a novel federated learning (FL)-enabled resource allocation optimization framework for e-health IoT systems. The framework leverages a three-level transmission architecture, including local training, wireless transmission, and global training levels. We model the delay and energy consumption for each level and formulate a joint optimization problem to minimize the long-term average system cost by jointly optimizing UE selection, transmission power, and computation resource allocation. Simulation results demonstrate the effectiveness of the proposed framework in reducing delay and energy consumption while maintaining privacy in e-health IoT systems.