Sequential Quadratic Method for GPS NLOS Positioning in Urban Canyon Environments

Abstract

In this paper, the problem of GPS non-line-of-sight (NLOS) positioning in urban canyon environments is considered. We propose a new position-determination estimator based on sequential quadratic programming (SQP) that is able to estimate and eliminate the path-delay error caused by the indirect transmission of the GPS signal. The estimator takes into account the measurement bias resulting from NLOS transmission and also improves the location accuracy of the satellite positioning system. The present method can effectively eliminate NLOS delay errors and improves the location accuracy of a satellite navigation receiver in an urban canyon environment. A Wilcoxon-norm-based regressor is further derived to improve the probability of detection of NLOS biases. The Wilcoxon regressor is a robust estimator that is well-suited to identifying outliers (in our case, NLOS biases) during the regression process. Experimental results demonstrate that the proposed estimator can accurately compute a user’s location after identifying and removing the measurement biases.