Robust Trajectory and Power Control for Cognitive UAV Secrecy Communication.

This paper investigates the physical layer security issue in an unmanned aerial vehicle (UAV) aided cognitive radio network. Specially, a UAV operates as an aerial secondary transmitter to serve a ground secondary receiver (SR) by sharing the licensed wireless spectrum assigned to primary terrestrial communication networks, and in the meantime multiple eavesdroppers (Eves) try to wiretap the legitimate UAV-to-SR link. Under the assumption that the location formation of the Eves is imperfect, we jointly optimize the robust trajectory and transmit power of the UAV over a finite flight period to maximize the SR's average worst-case secrecy rate, while controlling the co-channel interference imposed on the primary receivers (PRs) below a tolerable level. The design is formulated as a non-convex semi-infinite optimization problem that is challenging to be optimally solved. To deal with it, we first prove that the considered problem can be simplified as a more tractable one, which resolves the location uncertainties of the Eves without the aid of S-Procedure adopted in conventional methods. After that, an efficient iterative algorithm based on successive convex approximation (SCA) is developed to obtain a locally optimal solution. Numerical simulations are provided to demonstrate the effectiveness of our proposed algorithm and offer important system design insights.