Reinforcement Learning for Suppressing Eavesdroppers in Wireless Communication System

Abstract

The development of wireless communication technology requires higher security of wireless channel transmission. In wireless communication systems, active eavesdropping is a common method. Eavesdroppers send signals while eavesdropping, which provides the possibility for detection. Aiming at the problem of how to find the target position, this paper proposes a cooperative method to suppress eavesdropping nodes by using multi-agent method (SECM) in reinforcement learning. In our work, we introduce a UAV agent with a ranger of vision, which can find eavesdropping nodes through electromagnetic information and track their positions in real time. The UAV shares the eavesdropper’s position with the mobile jammer. Furthermore, fixed jammers are added in the scene to cooperate with the mobile jammer to pursue and form a cooperative strategy. The strategy can optimize the suppression efficiency. The simulation results show that the cooperative suppression eavesdropping algorithm proposed in this paper has better performance in detection than traditional Q-learning algorithm.