Satellite Navigation Signal Interference Detection and Machine Learning-Based Classification Techniques towards Product Implementation

Abstract

Many critical applications highly depend on Global Navigation Satellite Systems (GNSS) for precise and continuously available positioning and timing information. To warn a GNSS user that the signals are compromised, real-time interference detection is required. Additionally, real-time classification of the interference signal allows the user to select the most effective mitigation methods for the encountered disturbance. A compact proof of concept has been built using commercial off-the-shelf (COTS) components to analyse the jamming detection and classification techniques. It continuously monitors GNSS frequency bands and generates warnings to the user when interference is detected and classified. Various signal spectrum analyses, consisting of kurtosis and power spectral density (PSD) calculations, as well as a machine learning model, are used to detect and classify anomalies in the incoming signals. The system has been tested by making use of a COTS GNSS signal simulator. The simulator is used to generate the upper L-band GNSS signals and different types of interferences. Successful detection and classification is demonstrated, even for interference power levels that do not degrade the performance of a commercial reference receiver.