FL has surfaced as a notably distributed ML framework, where user devices collaboratively engage in the training of a shared ML model, supervised by a server. User devices in FL consecutively train local model updates (e.g., weight parameters or gradients) utilizing their proprietary data. Rather than transmitting raw, private data, user devices forward model updates to a server for amalgamation. In response, the server integrates local model updates to generate a comprehensive global model that is then dispatched to the devices for updating their respective local models [12, 23]. Such a communication cycle repeats until the model achieves a satisfactory accuracy level. FL prevents the potential unauthorized dissemination of private data [46]. For instance, FL allows multiple medical institutions to collaboratively train a unified ML model without directly sharing sensitive patient data.