Energy Efficiency Maximization for Aerial RIS-Aided Over-the-Air Federated Learning

Compared with terrestrial reconfigurable intelligent surface (RIS), aerial RIS provides improved transmission performance owing to its high adaptability in three-dimensional space and the existence of established line-of-sight links. In this paper, we study an aerial RIS-aided over-the-air federated learning (AirFL) system and propose an effective resource allocation scheme to maximize the energy efficiency (EE) of the AirFL system, where the transmit power of the Internet-of-Things devices, the receive scalar at the base station (BS), and the phase shifts of the RIS are jointly optimized. Since the formulated EE maximization problem consists of coupled variables, it is essentially non-convex. The original problem is then transformed into several sub-problems, which are addressed by adopting Dinkelbach's method, successive convex approximation and difference-of-convex programming approach. Finally, simulation results demonstrate that the proposed scheme is superior in terms of the EE performance for RIS-aided AirFL system compared with the existing scheme.