Reinforcement Learning Based UAV Trajectory and Power Control Against Jamming

Abstract

Unmanned aerial vehicles (UAVs) are vulnerable to jamming attacks that aim to interrupt the communications between the UAVs and ground nodes and to prevent the UAVs from completing their sensing duties. In this paper, we design a reinforcement learning based UAV trajectory and power control scheme against jamming attacks without knowing the ground node and jammer locations, the UAV channel model and jamming model. By evaluating the UAV transmission quality obtained from the feedback channel and the UAV channel condition, this scheme uses reinforcement learning to choose the UAV trajectory and transmit power based on the UAV location, signal-to-interference-and-noise ratio of the previous sensing data signal received by the ground node, and the radio channel state. Simulation results show that this scheme improves the quality of service of the UAV sensing duty given the required UAV waypoints and saves the UAV energy consumption.