Hybrid UAV-Enabled Secure Offloading via Deep Reinforcement Learning

Abstract

In this letter, we consider a secure offloading system consisting of a unmanned aerial vehicle (UAV)-mounted edge server, ground user equipments (UEs) and a malicious eavesdropper UAV. With the aim of maximizing secrecy sum-rate, we propose an adaptation of a helper UAV to switch the mode between jamming and relaying. We jointly optimize the helper UAV's trajectory and mode and UEs' offloading decision under energy budget constraints and operational limitations of nodes. The proposed algorithm is developed based on a deep deterministic policy gradient (DDPG)-based method, whose superior performances are verified via numerical results, as compared to other benchmark schemes.