Likelihood-based GNSS positioning using LOS/NLOS predictions from 3D mapping and pseudoranges

Abstract

The accuracy of conventional global navigation satellite systems (GNSS) positioning in dense urban areas is severely degraded due to blockage and reflection of the signals by the surrounding buildings. By using 3D mapping of the buildings to aid GNSS positioning, the accuracy can be substantially improved. However, positioning performance must be balanced against computational load. Here, a likelihood-based 3D-mapping-aided (3DMA) GNSS ranging algorithm is demonstrated that enables signals predicted to be non-line-of-sight (NLOS) to contribute to the position solution without explicitly computing the additional path delay due to NLOS reception, which is computationally expensive. Likelihoods for an array of candidate positions are computed based on the difference between the measured and predicted pseudoranges. However, a skewed distribution is assumed for those signals predicted to be NLOS on the basis that the ensuing ranging errors are always positive. An overall position solution is then extracted from the likelihood surface. GNSS measurement data have been collected at several locations in both traditional and modern dense urban environments. Horizontal root-mean-square single-epoch position accuracies of 4.7, 5.6 and 6.5 m are obtained using, respectively, a Leica Viva geodetic receiver, a u-blox EVK M8T consumer-grade receiver and a Nexus 9 tablet incorporating a smartphone GNSS antenna and a GNSS chipset that outputs pseudoranges. The corresponding accuracies using single-epoch conventional GNSS positioning are 20.5, 23.0 and 28.4 m, about a factor of four larger. The 3DMA GNSS algorithms have also been implemented in real time on a Raspberry Pi 3 at a 1-Hz update rate.