Generative adversarial network based resource optimization for UAV jamming assisted cognitive covert communication

Abstract

This paper discusses a resource optimization problem for unmanned aerial vehicle (UAV) jamming assisted cognitive covert communication. Specially, the secondary users adopt underlay strategy for spectrum access and the UAV transmits the noise signal to confuse the underlay detection of eavesdropper. According to the dynamic game characteristics in the covert communication process, a new generative adversarial network-based covert communication resource optimization algorithm (GAN-RO) is proposed for the joint optimization of UAV’s trajectory and power in cognitive covert communication assisted by UAV’s interference. The generator simulates the legitimate user to generate the resource allocation scheme, while the discriminator imitates the eavesdropper to detect covert communication. The dynamic game process between the generator and discriminator can obtain covert communication schemes by achieving Nash equilibrium. Simulation results indicate that the proposed GAN-RO algorithm can obtain a near-optimal resource optimization scheme with a rapid converge speed for non-ideal UAV jamming assisted cognitive covert communication system.