Federated Learning Framework with Lattice Joint Source-Channel Coding

The author introduces a novel federated learning framework using lattice joint source-channel coding to enhance over-the-air computation, focusing on quantizing model parameters and leveraging interference from devices.