Exploring Federated Learning: The Framework, Applications, Security & Privacy

Traditional machine learning models reveal short-comings in ensuring complete data security, leading to Federated Learning (FL) as a viable alternative, especially in emerging wireless network infrastructures such as Next Generation (NextG) or Open Radio Access Networks (O-RAN). The inclusion of FL in this process is important because centralized functionality facilitates collaborative learning without compromising the confidentiality of critical data. This review surveys the existing literature on FL, highlighting its basic principles, classification, potential applications, and approaches to various global models. Furthermore, it explores important issues that raise concerns about security and privacy in integrated learning and provides insights into potential avenues for research. Through rigorous analysis, this study highlights the importance of FL as a privacy protection mode of learning and considers its potential to shape the future of data-driven technology.