GNSS spoofing detection based on lightweight features and CGAN-ANN in unknown scenarios

The Global Navigation Satellite System (GNSS) with its open signal structure is highly vulnerable to covert spoofing interference, which limits the effectiveness of traditional detection methods in complex environments. To address the issues of feature redundancy and high computational complexity in existing multi-parameter detection methods, especially their insufficient detection capabilities in unknown scenarios, we propose a lightweight GNSS spoofing detector based on Conditional Generative Adversarial Network-Artificial Neural Network (CGAN-ANN). The proposed detector utilizes only a set of four core features extracted from the receiver’s Radio Frequency (RF) unit and tracking loop. By leveraging CGAN to generalize potential changes of multi-parameter feature, it effectively enhances the detection capability for spoofing signals in unknown scenarios. The results demonstrate that the proposed method achieves a detection rate exceeding 96% in all unknown scenarios, with an average detection rate of 98.58%. Compared to traditional and novel methods, the CGAN-ANN model improves the detection rate by approximately 8% in typical unknown scenarios and reduces detection latency by nearly 22 s. The proposed detector demonstrates excellent performance in terms of detection sensitivity, accuracy, and universality, providing an efficient and reliable anti-spoofing solution for GNSS receivers.