Trajectory Optimization for Multi-UAV Covert Communication Assisted by Non-Gaussian Noise

With the widespread application of unmanned aerial vehicles (UAVs) in areas such as cellular networks and disaster response, preventing their communication links from being externally detected and eavesdropped has become a challenge. In the context of covert secure transmission in a collaborative scenario of UAV nodes under non-Gaussian noise interference, we introduces a secure and efficient transmission design scheme based on Multi-Agent (MA) deep reinforcement learning. Initially, an optimization problem aiming to maximize secure covert transmission throughput is formulated and transformed into a MA Markov Decision Process (MDP). Subsequently, an Attention Mechanism (AM) combined with MA Twin Delayed Deep Deterministic Policy Gradient (MATD3) is employed to solve the problem using a MA deep reinforcement learning algorithm, optimizing the trajectories and power of UAV transmitters and jammers. Simulation results demonstrate that the proposed algorithm exhibits smoother convergence and achieves superior secure transmission performance in a multi-UAV network scenario assisted by non-Gaussian noise at UAV nodes.