Q-Learning Based UAV Secure Communication in Presence of Multiple UAV Active Eavesdroppers

This paper investigates UAV secure communication against multiple UAV active eavesdroppers (Eves). The active eavesdropper (Eve)can act adaptively as passive Eve or active jammer based on transmitting source power. In order to safeguard system security, we formulate the secure communication problem as a non-cooperative Stackelberg game and derive the equilibrium. Then Q-Learning algorithm is used to devise a Stackelberg equilibrium strategy in order to maximize the utility of the transmitter and optimize the secrecy rate. Simulation results indicate that the proposed scheme can enhance the security of the UAV communication in the presence of multiple UAV active Eves. Moreover, the impact of the distance between UAV Eves and legitimate transmitter-receiver on the secrecy rate and attack mode probabilities of Eves are also revealed.