A Novel Navigation Spoofing Algorithm for UAV Based on GPS/INS-Integrated Navigation

Although covert navigation deception technique entails high computational complexity, its high concealment has garnered increasing attention. However, existing research typically relies on known unmanned aerial vehicle (UAV) reference trajectories and other internal filtering parameters to calculate deceptive positions, which are hard to obtain in reality. In this paper, addressing the condition where internal Kalman filtering (KF) parameters and preset reference trajectory parameters of the victim UAV cannot be observed, a novel method for intelligent navigation deception position estimation based solely on radar detection of UAV motion information is proposed. This method essentially exposes the nonlinear mapping relationship between navigation deception positions and autonomous UAV motion control, while also considering the Markovian nature of the navigation deception process. Deep reinforcement learning (DRL) technique is introduced to dynamically solve for effective navigation deception positions in real-time. To ensure the generated navigation deception positions meet concealment requirements, a concept of deception residual rate based on radar estimation is proposed, with its impact on concealment and stability theoretically analyzed. Subsequently, we design a reward function that simultaneously considers navigation deception position concealment, deception trajectory stability, and the successful enticing of the victim UAV to the desired destination. Comparison experiments demonstrate that the estimated deception position by the proposed method achieves similar concealment and stability effects comparable to previous algorithms in the absence of reference trajectories and KF parameters, and successfully deceives UAVs to the intended destination while remaining below the normalization innovation square (NIS) threshold of UAV deception detection.