Evaluation of (Un-)Supervised Machine Learning Methods for GNSS Interference Classification with Real-World Data Discrepancies
🤖 AI Summary
To address unreliable GNSS positioning in vehicular systems caused by real-world interference, this study systematically evaluates supervised (CNN, LSTM, Transformer) and pseudo-label-based unsupervised learning for interference signal classification, using large-scale real-world measurements from German highways and Austrian Alpine roads. Its key contributions include: (i) the first empirical validation of pseudo-labeling unsupervised learning in large-scale realistic vehicular GNSS scenarios; (ii) identification of cross-regional environmental discrepancies as a critical constraint on model generalization; and (iii) proposal of a synergistic adaptation framework integrating anomaly detection (Isolation Forest), domain adaptation (DANN), and time-frequency-domain data augmentation. Experimental results show a classification accuracy of 98.2%, with pseudo-labeling achieving 92% of supervised method performance; DANN improves cross-scenario F1-score by 37%, significantly mitigating data distribution shift.
Application Category
📝 Abstract
The accuracy and reliability of vehicle localization on roads are crucial for applications such as self-driving cars, toll systems, and digital tachographs. To achieve accurate positioning, vehicles typically use global navigation satellite system (GNSS) receivers to validate their absolute positions. However, GNSS-based positioning can be compromised by interference signals, necessitating the identification, classification, determination of purpose, and localization of such interference to mitigate or eliminate it. Recent approaches based on machine learning (ML) have shown superior performance in monitoring interference. However, their feasibility in real-world applications and environments has yet to be assessed. Effective implementation of ML techniques requires training datasets that incorporate realistic interference signals, including real-world noise and potential multipath effects that may occur between transmitter, receiver, and satellite in the operational area. Additionally, these datasets require reference labels. Creating such datasets is often challenging due to legal restrictions, as causing interference to GNSS sources is strictly prohibited. Consequently, the performance of ML-based methods in practical applications remains unclear. To address this gap, we describe a series of large-scale measurement campaigns conducted in real-world settings at two highway locations in Germany and the Seetal Alps in Austria, and in large-scale controlled indoor environments. We evaluate the latest supervised ML-based methods to report on their performance in real-world settings and present the applicability of pseudo-labeling for unsupervised learning. We demonstrate the challenges of combining datasets due to data discrepancies and evaluate outlier detection, domain adaptation, and data augmentation techniques to present the models' capabilities to adapt to changes in the datasets.
Problem
Research questions and friction points this paper is trying to address.
Evaluates ML methods for GNSS interference classification with real-world data
Assesses feasibility of ML in real-world GNSS interference scenarios
Explores dataset challenges and adaptation techniques for GNSS interference
Innovation
Methods, ideas, or system contributions that make the work stand out.
Supervised ML methods for GNSS interference classification
Pseudo-labeling for unsupervised learning adaptation
Domain adaptation and data augmentation techniques
🔎 Similar Papers
2024-09-23arXiv.orgCitations: 2
💼 Related Jobs
No related jobs found.
