Anti-Eavesdropping in UAV Secure Communications: Joint Trajectory and Power Allocation Optimization with Two-Way Training

Abstract

This paper investigates an unmanned ariel vehicle (UAV)-enabled secure communication system, where an UAV sends confidential information to legitimate ground users (LGUs) in the presence of a ground eavesdropper (Eve). In order to improve the comprehensive security performance of UAV network communication, a two-way training-based discriminatory channel estimation (DCE) scheme focusing on physical layer security (PLS) is proposed. We derive the closed-form expression of the average secrecy rate (ASR) and the minimum normalized mean square error (NMSE). On the premise of ensuring the effectiveness of channel estimation errors, we aim to maximize the ASR of the system, by jointly optimizing the UAV's flight trajectory and power allocation. Due to the intractability of the formulated problem, block coordinate descent (BCD) and successive convex approximation (SCA) techniques are applied to solve the problem. Numerical results show that the proposed scheme can enhance the security of the UAV communication. Moreover, the relationship between system ASR and UAV power allocation is revealed.