Energy-Efficient Secure Federated Learning for UAV Swarms

Unmanned Aerial Vehicles (UAVs) have been recently leveraged in massive amount of Internet of Things (IoT) applications. However, given the stringent limitations of UAVs, investigating their performance in terms of their energy efficiency, tasks accuracy, and security has gained much interest recently. In this paper, we study the deployment of Federated Learning (FL) within the UAV swarms, with the aim of avoiding data exchange between different swarm's members. Specifically, we propose a secure FL framework for UAV swarms, such that the energy efficiency is maximized, while enhancing the learning performance. The suggested approach takes into account the issue of data imbalance in FL in addition to the strict energy and security limits placed on the UAVs.Our findings show how effective our framework is in terms of security, energy efficiency, and classification accuracy when compared to baseline systems that don't take security or data imbalance into account.