Jammer Classification in GNSS Bands Via Machine Learning Algorithms.

Area of Science:

• Signal Processing
• Machine Learning
• Navigation Systems

Background:

• Global Navigation Satellite System (GNSS) signals are vulnerable to jamming, necessitating robust detection and classification methods.
• Existing jammer classification techniques may lack efficiency or accuracy in diverse jamming scenarios.

Purpose of the Study:

• To develop a novel approach for classifying jammers in GNSS bands by reframing the problem as image classification.
• To evaluate the performance of machine learning algorithms, specifically Support Vector Machines (SVM) and Convolutional Neural Networks (CNN), for jammer classification.

Main Methods:

• Time-frequency analysis of jammed GNSS signals to generate image representations.
• Application of machine learning algorithms (SVM and CNN) for classifying signals into six categories (jammer absent or five jammer types).
• Creation and provision of open-access training and test datasets for jammer classification.

Main Results:

• Support Vector Machines (SVM) achieved a classification accuracy of up to 94.90%.
• Convolutional Neural Networks (CNN) demonstrated a classification accuracy of up to 91.36%.
• The proposed image classification method effectively distinguishes between the absence of a jammer and five different jammer types.

Conclusions:

• Treating GNSS jammer classification as an image recognition problem is a viable and effective strategy.
• Machine learning, particularly SVM, offers high accuracy for classifying GNSS jamming signals.
• The open-access datasets will facilitate further research and development in GNSS anti-jamming techniques.