Optimized algorithm for BDS/GPS RTK suitable for urban canyons with a low-cost receiver

Abstract

Due to large noise, severe multi-path errors and the frequent rise and fall of satellites caused by propagation obstacles and mirror materials widespread in urban canyons with low-cost receivers produced with low-power design and restricted computing resources, the positioning accuracy in urban canyons with normal RTK has been heavily reduced with these large biased observations and discontinuous tracking satellites. Therefore it is not applicable to urban canyons without optimization. The C/N0-dependent exponential noise model for an urban canyon was designed to provide a balance between practical noises and the noise model adopted in the EKF filter. The accuracy of the fixed solution with this model was better than 0.5 m compared to the traditional elevation-dependent model for urban canyons, except for special circumstances, such as a deep urban canyon. So, an optimized empirical strategy for ambiguity resolution depending on posterior variance and the number of satellites available was also proposed for special local circumstances, such as a deep urban canyon. The results show that it has a heavily reduced proportion of incorrectly-fixed solutions; the accuracy of the fixed solution can be better than 0.10 m with a significant drop in incorrectly-fixed solutions, and the accuracy of the float solution can be better than 0.898 m with good reliability and stability, even if the ambiguity resolution failed. The proportion of correct-fixed solutions in a deep urban canyon was further reduced with the ambiguity resolution AR-AIM method, which was proposed to migrate the negative effect of the frequent rise of satellites on the ambiguity resolution.