GIS based robust GNSS positioning technique for vehicular navigation in urban environments

Abstract

A GIS (Geographic Information System) based positioning technique is defined as a positioning technique using geographic information in order to improve the positioning accuracy. When vehicles equipped with a GPS (Global Positioning System) receiver are driving in urban environments, the re-ceiver may have considerable positioning error due to multipath and non-line-of-sight (NLOS) condi-tions. We suggest algorithms to solve this problem. We propose two algorithms to improve low-cost vehicle GNSS modules with DGPS service. First, we suggests the Pseudorange Inverse Estimation (PIE) algorithm to restore measurement directly without requiring raw pseudorange data. On the other hand, the proposed PIE algorithm can be used with com-mercial vehicle GNSS receivers that provide National Marine Electronics Association (NMEA) format data. Second, we suggest GIS based VSC-WLS (Virtual Satellite Constrained - Weighted Least Square) techniques to correct errors in urban environments in particular. We build a 3D building model made with GIS data, and use it to determine whether the received signals are LOS or NLOS. We then place a virtual satellite (VS) obtained through the stochastic calculation using SNR (Signal to Noise Ratio) measurement model. Finally, we obtain satisfactory positioning results using WLS (Weighted Least Square) estimation with the VS. It complements low-cost vehicle GNSS positioning and can significantly improve the cross-street positioning accuracy in dense urban environments. This paper describes how proposed algorithms have been adapted to work on a smartphone and presents the comprehensive per-formance assessment of low-cost vehicle GNSS positioning. Using GPS and GLONASS data recorded at 12 locations within Gangnam Teheranro, it is shown that VSC-WLS techniques significantly outper-forms conventional GNSS positioning in the cross-street direction.