Content Caching in HAP-Assisted Multi-UAV Networks Using Hierarchical Federated Learning

Abstract

High and low altitude platforms are expected to become an important component in current access infrastructure design to improve the radio access capability and support on-demand edge services. Caching popular contents at edge such as unmanned aerial vehicles (UAVs), can meet the requirements of mobile users that have the same content requirements, without duplicate transmissions through the backhaul links. Therefore, content access delay can be significantly reduced. In this paper, we propose an intelligent and collaborative popularity prediction method for content caching in high altitude platform (HAP)-assisted multi-UAV networks supported with a hierarchical federated learning algorithm. In this way, the proposed method also preserves the privacy of the contents of mobile users. Simulation results show that the proposed method achieves good prediction accuracy by reducing the prediction error significantly.