Completion Time Minimization of Fog-RAN-Assisted Federated Learning With Rate-Splitting Transmission

This work studies federated learning (FL) over a fog radio access network, in which multiple internet-of-things (IoT) devices cooperatively learn shared machine model by communicating with cloud server (CS) through distributed points (APs). Under the assumption that fronthaul links connecting APs to CS have finite capacity, rate-splitting transmission at IoT (IDs) is proposed enables hybrid edge and decoding of split uplink messages. The problem completion time minimization for FL tackled optimizing quantization strategies along training hyperparameters such as precision iteration numbers. Numerical results show achieves notable gains benchmark schemes rely solely on or decoding.