A Partial Federated Learning Model in Cognitive UAV-enabled Edge Computing Networks

Abstract

Modern wireless networks need to incorporate fed-erated learning (FL) with communications to minimize latency and protect privacy. The latency issue with Internet-of-things devices can be resolved by using FL over wireless networks. In this paper, we present a cognitive network of unmanned aerial vehicles that aims to reliably offer edge computing services to Internet-of-things devices in a specific location. To reduce the latency of Internet-of-things devices, we design a partial FL model implemented in unmanned aerial vehicles. Simulation findings show that the partial FL model can outperform the FL model by 21.72%.