Innovating Asynchronous Federated Learning for High-Latency Environments

This research explores the limitations of synchronous federated learning in high-latency settings. We propose an asynchronous framework and validate its performance through simulation experiments, focusing on Earth-Mars collaborative missions, addressing delays, data heterogeneity, and computational constraints to enhance interstellar communication efficiency.