Elusion-Based UAV-Assisted Covert Data Collection Under Uncertain Eavesdropper Mobility

This paper studies how to enable efficient covert data collection in a UAV-assisted network under the surveillance of an aerial eavesdropper. Unlike existing work, we consider a mobile eavesdropping UAV which patrols the area of interest with an unknown mobility pattern. By exploiting the legitimate UAV’s mobility, we propose an elusion-based covert data collection scheme by optimizing the trajectory of the legitimate UAV so that it always performs data collection at the locations far away from the eavesdropper to improve the covert throughput. Given the uncertain mobility of the eavesdropper, we formulate the design of the proposed data collection scheme as a reinforcement learning problem and develop a covertness-aware learning algorithm to facilitate efficient solution search. Simulation results demonstrate that, in comparison to existing proposals, our proposed scheme can effectively improve the performance of covert data collection.