A Robust GNSS LOS/Multipath Signal Classifier based on the Fusion of Information and Machine Learning for Intelligent Transportation Systems

In urban canyons, transportation systems localization based on GNSS (Global Navigation Satellite System) can be strongly affected, due to NLOS and Multipath effects. The GNSS signal can be reflected by the obstacles near of the vehicle (buildings, buses, Trees...), and these reflections degrade significantly the positioning accuracy. To improve positioning performance of intelligent transportation systems in harsh environments, several solutions in the literature have shown that a good signal classification can significantly improve positioning accuracy in signal degradation environments but is limited by low reliability. In this research work, we address a very challenging problem related to GNSS signals reception state detection, classified as a Decision problem. A new GNSS signal classifier system based on the fusion of information provided by RHCP and LHCP antennas, and a machine learning technique is proposed. The novelty of our proposed classifier summarizes in exploiting the characteristics and potential of the RHCP and LHCP antennas to process the GNSS signal (a new GNSS receiver is proposed). Furthermore, using together a machine learning technique, namely, Decision Trees algorithm, and data fusion is a promising approach for improving data analysis and classification, especially in complex scenarios. The proposed DecisionTrees-based classifier has used the satellite elevation and C/N0-R-L ratio, which is the difference between the left-hand circular polarized (LHCP) C/N0 (Carrier-to-noise) and the right-hand circular polarized (RHCP) C/N0.Effective and proper data classification by applying the proposed classifier is demonstrated through several field tests in urban areas, Calais, France. We obtain an accuracy of 99% over KNN and SVM classification techniques.