A deep learning GNSS spoofing and jamming detection method with dual-frequency C/N0 heatmap

Abstract

Abstract The Global Navigation Satellite System (GNSS) is vulnerable to interference due to the open signal structure and low signal strength, posing a significant threat to the billions of terminals worldwide that rely on GNSS receivers for precise Positioning, Navigation, and Timing (PNT) services. In this paper, we propose a cloud-edge framwork for GNSS spoofing and jamming monitoring, comprising the data acquisition module, GNSS monitoring module, detecting and reporting module. In this framwork, we design a Deep Learning (DL) method for detecting GNSS interference through Dual-frequency Carrier-to-Noise density ratio (C/N0) heatmaps (DD-C/N0). This method involves extracting and correlating features from C/N0 heatmaps of visible navigation satellites operating in the GPS L1 and L2 frequency bands, allowing the identification of anomalous patterns. A U-BLOX receiver was utilized to capture the GNSS satellite signals, while commercial jammers and Software-Defined Radio (SDR) HackRF One kits were employed to simulate the interference sources. Experimental results demonstrate that the proposed method achieves significantly higher performance, with an accuracy of 99% and 98% on the public dataset and real-time testing data, compared to unsupervised, semi-supervised, and supervised detectors that rely solely on single-channel data (L1 frequency band). Integrated with the DD-C/N0 method, the online GNSS monitoring system will be improved and deployed to automate spoofing and jamming detection tasks in the next step.