Abstract: The issue of data heterogeneity has long been a significant challenge in federated learning, impacting the speed of global model convergence and potentially lowering the overall performance ...