I. Introduction
Federated learning (FL) holds significant importance in the realm of machine learning and data privacy via decentralized learning approaches [1]. It allows model training across multiple devices or servers without needing to centralize raw data, thus preserving user privacy and data security. As models could be trained locally on user devices and only transmitting model updates instead of raw data, federated learning minimizes the risk of data breaches and ensures that sensitive information remains on users’ devices. This approach not only enhances privacy but also facilitates the development of more robust and personalized machine learning models by leveraging diverse data sources. Consequently, federated learning paves the way for advancements in various fields such as healthcare, finance, and telecommunications, where data privacy is paramount. However, it is still facing some practical challenges such as expensive communication, systems/statistical heterogeneity and privacy concerns [2].