Single-Source and Distributed GNSS Interference Classification based on Spatial Cross-Correlation Spectrum and CNN

Yan Huang, Chun Wang, Yuxuan Huang, Xiang Chen

Peer Reviewed

Abstract:	The classification of interference signals and the estimation of direction of Arrival (DOA) are challenges in the anti-spoofing of Global Navigation Satellite Systems (GNSS), especially as the interference scenario is becoming complex. In-band interferences in navigation receiving environments include jamming and spoofing. The jammer emits high-power signals, drowning out the authentic signals in the frequency domain and causing the receiver to lose lock. But spoofing spreads false navigation information. According to their degree of synchronization with the authentic signals, they are mainly divided into two situations: asynchronous spoofing and synchronous spoofing. From the perspective of attack methods, spoofing attacks can be further classified into single-source attacks and distributed attacks. Distributed spoofings are more concealed by simulating multiple satellites to form forged signals that are transmitted in different directions. Perceiving interference scenarios from all directions of the environment is a prerequisite for making wise anti-interference decisions. Therefore, in this paper, a method named SCCS-CNN is proposed by combining array processing and deep learning. This method combines the spatial correlation spectrum (SCCS) with the lightweight convolutional neural network (CNN)- ResNet18, which has strong image classification performance. Given the limitations of time-frequency domain analysis in detecting spoofing, the highly correlated characteristics existing in the spatial domain offer a new option. This method utilizes the multi-dimensional characteristics of array antennas receiving signals combined with SCCS capturing interference signals, including DOA, power, and cross-correlation information. By using ResNet18 to extract features from SCCS, six different scenarios can be identified: jamming, single synchronous/asynchronous spoofing, distributed synchronous/asynchronous spoofings, and a non-interference scenario. The experimental results show that the SCCS-CNN classification method combined with spatial domain features can effectively identify new distributed synchronous spoofings. While providing the classification results, this method also outputs multidimensional features of the interference, thereby providing valuable information for subsequent decision suppression measures.
Published in:	Proceedings of the 2026 International Technical Meeting of The Institute of Navigation January 26 - 29, 2026 Hyatt Regency Orange County Anaheim, California
Pages:	454 - 468
Cite this article:	Huang, Yan, Wang, Chun, Huang, Yuxuan, Chen, Xiang, "Single-Source and Distributed GNSS Interference Classification based on Spatial Cross-Correlation Spectrum and CNN," Proceedings of the 2026 International Technical Meeting of The Institute of Navigation, Anaheim, California, January 2026, pp. 454-468. https://doi.org/10.33012/2026.20506
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In