Federated learning: analysis, applications and perspectives

In the last few years, a lot of devices such as mobile phones, are equipped with progressively sophisticated sensing and computing capabilities. Typical cloud-based Machine Learning (ML) methods require the data to be centralized in a cloud server or data center. However, this outcome in critical concerns is associated with unbearable latency and communication inefficiency. For this purpose, Mobile Edge Computing (MEC) has been proposed to bring intelligence closer to the edge, where data is produced. Nevertheless, the common permitting technologies for ML at mobile edge networks still require personal data to be shared with outside parties, e.g., near edge servers. Lately, considering increasingly harsh data privacy legislations and growing privacy issues, the concept of Federated Learning (FL) has been introduced. In FL, end devices use their own local data to train an ML model required by the server. Then they send the model updates rather than raw data to the server for aggregation. FL can serve as enabling technology in mobile edge networks since it enables the collaborative and decentralized training of an ML model and moreover authorizes DL for mobile edge network optimization. Then, we highlight the aforesaid challenges of FL implementation and application. Furthermore, we present the applications of FL for end devices and mobile edge network optimization through the Flower Framework. Finally, we discuss the results, the important challenges, and future research directions in FL