Detecting Multipath Effects of GNSS by Using Robust Estimator

In this dissertation, a new positioning method based on the robust estimation (R-estimation) is proposed to detect the multipath effect in GNSS/GPS system. We reorganize the GNSS/GPS positioning problem as a nonlinear regression problem and point out the positioning result of the traditional least squares approach is sensitive to the pseudorange measurement outliers caused by the non-line-of-sight (NLOS) signal. The new rank-based approach from the R-estimator is adopted here to provide better robustness.

The Wilcoxon norm is used as a new objective function in this new approach. The numerical method, called the gradient decent method, is provided to minimize the Wilcoxon objective function. The properties of this new objective function will be thoroughly discussed in our dissertation by analyzing 2-dimension cases. Comparing with the 2-norm objective function, the new Wilcoxon objective function will not be changed with the variation of the receiver clock bias.

The timing of applying the rank-based approach will be discussed in terms of the standard deviation of the residuals with respect to the minimized 2-norm objective function. The performance of detecting the satellites corrupted by the multipath effect will be provided through the simulation of dual GNSS constellations. The results show the proposed approach has better performance against larger outliers.