MATCHING COMBINED MULTI-AGENT REINFORCEMENT LEARNING FOR UAV SECURE DATA DISSEMINATION

Due to the high flexibility and mobility, unmanned aerial vehicles (UAVs) can be deployed as aerial relays touring to disseminate data to ground users (GUs), especially when the ground base station is temporally dysfunctional or damaged. However, the broadcasting nature of wireless communication leads to the security issue with the presence of malicious eavesdroppers (Eves). The paper proposes a matching combined multi-agent deep reinforcement learning (DRL) to maximize the average secure rate for UAV-ground communications in probabilistic line-of-sight (LoS) channels, with the joint consideration of UAVs' propulsion energy and trajectory, as well as GUs' dissemination data size. The numerical simulation demonstrates that comparing with the other no matching combined DRL or valued-based DRL approaches, the proposed matching combined multi-agent deep deterministic policy gradient (matching-MADDPG) has better performance at both trajectory and convergence.