Design and Evaluation of Robust M-estimators for GNSS Positioning in Urban Environments

Abstract

Providing a robust position solution in urban scenarios using Global Navigation Satellites Systems (GNSS) supposes a big challenge due to the continuous presence of local threats like multipath, non-light-of-sight signals (NLOS) or interference. The impact of these threats in the pseudorange measurements is difficult to model due to the constant changing environment. Therefore they have to be dealt with at position estimator level. In this paper we leverage the knowledge from robust statistics and we design and apply Huber M-estimators to the positioning and compare its performance with respect to the classical least-squares estimator. We provide further insights on the importance of the chosen suitable pseudorange error models for the robust estimator process by processing real GNSS data in open-sky and deep urban scenarios. Under a proper design, these estimators have the potential to became a powerful option in challenging GNSS scenarios under the presence of multiple corrupted measurements or data that do not follow the assumed error models.