Real-Time GPS Spoofing Detection in Consumer Drones Through Multi-Sensor Data Fusion and TimesNet

To address the threat of GPS spoofing attacks on consumer drones, this paper proposes a real-time detection method based on multi-sensor data fusion and TimesNet. By integrating data from multi-source heterogeneous sensors such as the Inertial Measurement Unit (IMU), magnetometer, and barometer, a spatiotemporal feature cross-validation mechanism is constructed, overcoming the limitations of traditional single-sensor detection. Innovatively, the TimesNet deep temporal network is introduced, leveraging key frequency-domain period extraction and multi-scale feature reorganization techniques to achieve the second-level real-time detection response. A dynamic adversarial GPS spoofing sample generation algorithm is proposed to simulate multi-dimensional position offset attacks in complex battlefield environments, constructing an adversarial sample library covering eight typical flight trajectories. Experimental results demonstrate that the proposed method achieves an average detection accuracy of 99.57% and an F1-score of 99.30% across complex trajectories such as straight lines, curves, and spirals, with a detection response time controlled within 0.35-0.70 seconds. Compared to existing visual matching and single-sensor solutions, this method requires no additional hardware and can be deployed on embedded platforms with only 1.2M parameters, providing a lightweight solution for drone navigation security.