A Label-Free Framework for GNSS Spoofing and Jamming Detection With Cluster-Level Semantic Inference

IEEE Communications Letters 2026

View at publisher
Authors: Qi Yao, Jingjing Yang, Shunfang Wang, Zhe Xiao, Ming Huang
DOI: 10.1109/lcomm.2026.3670467
Cited by: 1
References: 14

Abstract

Global Navigation Satellite System (GNSS) spoofing and jamming have become increasingly prevalent, while real-world attacks are often abrupt, short-lived, and difficult to annotate. Motivated by these challenges, this letter designs a label-free detection framework. An autoencoder and unsupervised clustering extract latent structures from unlabeled GNSS data. A large language model evaluates cluster-level statistics, while a Gaussian mixture model (GMM) characterizes the latent-space distribution. Their outputs are combined to produce anomaly scores without labeled data. Experiments on real-world datasets achieve 93.27%/95.44% AUC/accuracy for spoofing and 97.45%/97.1% for jamming.

Fields of Study & Subjects

Indoor and Outdoor Localization Technologies GNSS positioning and interference Anomaly Detection Techniques and Applications

Publication details

Journal / source
IEEE Communications Letters
Publisher
Institute of Electrical and Electronics Engineers (IEEE)
Volume
30
Pages
1280–1284
Publication date
1 January 2026