Joint Jamming and Transmission Rate Optimization for Multi-Jammer Assisted UAV Covert Communications With Imperfect CSI

Information security and privacy are critical for unmanned aerial vehicles (UAV) based low-altitude networks. To enhance the security of the UAV transmitter, this paper inves tigates the jammer-assisted UAV covert communication systems with imperfect CSI, where part of friendly jammers is selected to send artificial noise (AN) to confuse the warden. Then, the UAV transmitter can hide the transmitted signals into “noise”, decreasing warden's detection probability for the transmission behavior. It is important to jointly optimize the jammer selection and jamming power, as AN sent by friendly jammers will interfere with the warden and receiver, simultaneously. However, existing channel-aware schemes work with a perfect channel state information (CSI) of the transmitter-to-receiver (i.e., covert link) and jammer-to-receiver links, which is impractical in low-altitude networks. Therefore, this paper concerns the imperfect CSI of the covert link and the channel distribution information (CDI) of the UAV transmitter-to-warden link and jamming links from the jammers to the warden and receiver. First, given the imperfect CSI, the warden's average detection error probability (WADEP) is analyzed. Then, given that WADEP is higher than the covert ness constraint, the covert throughput is analytically derived and maximized by jointly optimizing the jammer selection, jamming power, and transmission rate (J-JamSPR-ImCSI). Since the optimization problem is NP-hard, a two-step Jensen's inequality based algorithm is proposed to provide a heuristic solution (2 Step-JIHA). The main idea is to first jointly optimize jammer selection and jamming power with the covertness constraint to minimize the average received AN power at the receiver (which negatively impacts the covert throughput), then a bisection method can be employed to obtain the optimal transmission rate maximizing the covert throughput. Numerical results show that the proposed 2-Step-JIHA can provide a near optimal covert throughput with a lower complexity. Moreover, when the CSI is imperfect, compared to the existing channel-aware scheme, the proposed J-JamSPR-ImCSI with 2-Step-JIHA can improve the covert throughput by 54.26%, when the covertness constraint is 0.9.