Urban environment recognition based on the GNSS signal characteristics

Statistical characteristics of signal reception conditions vary greatly in different types of environments. Hence, Global National Satellite System (GNSS) receivers must recognize surroundings for choosing the most suitable positioning methods in real time. Targeting vehicular positioning applications in a city, a novel environment recognition algorithm based only on the GNSS signal characteristics is proposed to distinguish between six distinct settings. To characterize different environmental interferences, a signal feature vector is built to represent the signal attenuation, blockage, and multipath. By training the classification model with labeled feature vectors, the support vector machine (SVM) algorithm is used to predict the scene type. A temporal filtering method is proposed to improve the accuracy. With advanced training of the model, this recognition method can work for the receiver in real time. To prove the extensive applicability of the proposed algorithm, the prediction data set and the training data set are collected in different cities. The testing results show overall recognition accuracy of 89.3% across different environments.