A Machine Learning Based GNSS Signal Classification

The global navigation satellite system (GNSS) provides accurate position, velocity, and time data all over the world. However, GNSS are susceptible to multipath effects in suburban, urban, and indoor environments. Furthermore, it is vulnerable to intentional interference such as jamming and spoofing, which can result in either no or false position estimates. This study focuses on classifying received GNSS signals in a multi-correlation GNSS receiver as interference-free, multipath, jamming, or spoofing. To classify the GNSS signal, the average power and distortion correlation features are extracted from the multi-correlation output. Various machine learning algorithms such as neural networks, support vector machines, nearest neighbors, kernel approxi-mation, decision trees, discriminant analysis, naive Bayes, and ensemble classifiers are investigated and quantified using test accuracy and confusion matrix.