Weighted Centroid GNSS Interference Source Localization Based on Unsupervised Neural Networks

Lixin Zhang, Renbiao Wu, and QiongQiong Jia

Peer Reviewed

Abstract:	The inherent vulnerability of the Global Navigation Satellite System (GNSS) makes it susceptible to various interference signals. Accurate localization of interference sources can quickly mitigate interference, ensuring the stable operation of GNSS. The Received Signal Strength (RSS) based localization method is widely used due to its simplicity; however, it is sensitive to environmental factors such as building occlusion and to the distribution of observation points. To overcome these limitations, this paper proposes a weighted centroid localization method based on an unsupervised neural network, with the system architecture built on an Unmanned Aerial Vehicle (UAV) platform. First, the UAV collects satellite signal Carrier to Noise Ratio (C/N0) measurements in the air, effectively avoiding or reducing the blockage and reflection of satellite signals caused by ground buildings or obstacles. The neural network is then used to estimate the weights of different observation points, which are combined with the weighted centroid method to localize the interference source, enhancing adaptability in cases of sparse or unevenly distributed observation points. A ranking-entropy loss function is designed for unsupervised training of the neural network, improving robustness in various environments. Experimental results show that, even when the interference source lies outside the coverage area of the observation points, the localization error is only 122.27 meters, which is at least 409 meters lower than that of other methods.
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:	516 - 526
Cite this article:	Zhang, Lixin, Wu, Renbiao, Jia, QiongQiong, "Weighted Centroid GNSS Interference Source Localization Based on Unsupervised Neural Networks," Proceedings of the 2026 International Technical Meeting of The Institute of Navigation, Anaheim, California, January 2026, pp. 516-526. https://doi.org/10.33012/2026.20528
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In