Aerial Access Networks for Federated Learning: Applications and Challenges
Quoc‐Viet Pham, Ming Zeng, Thien Huynh‐The, Zhu Han, Won–Joo Hwang
Abstract
Aerial access networks (AANs) and mobile edge computing (MEC) have been considered as key enablers of future networks. In this article, we investigate the application of MEC-empowered AANs (also known as aerial computing) for federated learning (FL), a promising technology for providing private and distributed solutions to mobile edge networks. We first introduce the fundamentals of AANs and FL, and illustrate the potential benefits of aerial FL networks. On this basis, we present important applications of AANs for FL. It is shown that distinctive characteristics such as flexible deployment and high mobility, when exploited cleverly, can provide various benefits for FL-enabled networks. Finally, major challenges and potential directions are highlighted.