Robust Two-Layer UAV Spoofing Detection Based on Multi-Source/Multi-Dimensional Data Analysis for Low-Altitude Networks

The navigation safety of Unmanned Aerial Vehicles (UAVs) is crucial for the stability and reliability of low-altitude networks. However, the Global Navigation Satellite System (GNSS) they depend on is vulnerable to spoofing, becoming a key challenge to low-altitude network security. Existing detection methods typically focus on single attack types or single-sensor data, which limits effectiveness against hybrid spoofing. Therefore, we propose a multi-source, multi-dimensional data analysis-based spoofing detection model, combining complementarity of GNSS and Inertial Measurement Unit (IMU) to conduct a two-layer detection. By employing a sliding window mechanism, the first layer performs rapid threshold detection by leveraging statistical features differences between spoofed and genuine signal data, while the second layer conducts a refined position-attitude consistency check if the first layer reports no spoofing. When no spoofing is detected by two layers, GNSS data is utilized to correct IMU drift, forming a closed-loop “detection-correction” system that enhances long term consistency. Monte Carlo simulations demonstrate that the proposed method effectively detects both step and ramp hybrid spoofing attacks. Under attacks with a maximum spoofing offset of$d_\text{max}=9m$, the detection probability exceeds 95%, false alarm rate remains below 3%, and average detection time is reduced by 57% compared to direct-comparison and Kalman filter innovation-based approaches.