A Q-learning based Method f or Secure UAV Communication against Malicious Eavesdropping

Abstract

Due to the advantages such as excellent mobility, low-cost, on-demand deployment, unmanned aerial vehicles (UAVs) are expected to play a significant role in the future wireless communication systems. One of the necessary and urgent problems is the physical layer security of this wireless communication system. Specifically, under the open nature of the UAV wireless channels, how to convey confidential information reliably in a coexistence environment with eavesdroppers. In this paper, we investigate a Q-learning based security scheme to maximize the average secrecy rate (ASR) against intentional or unintentional eavesdropping. Simulation results show comparable disguising performances compared with the benchmark method.