Abstract & Overview
Insufficient localization accuracy of global navigation satellite system (GNSS) receivers is one of the challenges to implement advanced intelligent transportation system in highly urbanized areas. Multipath and non-line-of-sight (NLOS) effects strongly deteriorate GNSS positioning performance. This paper aims to train a classifier by supervised machine learning to separate the type of GNSS pseudorange measurement into three categories, clean, multipath and NLOS. Several features obtained or calculated from the GNSS raw data are evaluated. This paper also proposes a new feature to indicate the consistency between measurements of pseudorange and Doppler shift. According to the experiment result, about 75% of classification accuracy can be achieved using a support vector machine (SVM) classifier trained by the proposed feature and received signal strength.