A weighted GPS positioning algorithm for urban canyons using dual-polarised antennae

Abstract

The positioning performance of the Global Positioning System (GPS) is severely degraded in urban canyons due to buildings blocking or attenuating signals. This paper proposes a weighted GPS positioning method using a dual-polarisation antenna in order to deal with this issue. The GPS signals are first classified and labelled into NLOS signals and LOS/MP signals by considering the right-hand circular polarised (RHCP) and the left-hand circular polarised (LHCP) signal strengths. Then, a Bayesian optimisation-based Gaussian process is used to fit the pseudorange error related to the input of the carrier-to-noise ratio of the RHCP QUOTE $\mathrm{C}/{{\mathrm{N}}_0}^{\mathrm{R}}$( $C/N_0^R$) and the elevation angle of the LOS/MP signals. The pseudorange errors of the LOS/MP signals are therefore predicted based on the rules extracted from the fitting results. The positioning solutions are then obtained from the weighted least-squares algorithm, with the weighting strategies being based on the predicted pseudorange errors. Static positioning results in deep urban areas showed that the horizontal root-mean-square error (RMSE) positioning has been improved by 64.6% and 57.3%, respectively, while 3D RMSEwas improved by 74.3% and 48.1%, respectively, compared with traditional single-point positioning results.