Novel Snapshot Integrity Algorithm for Automotive Applications: Test Results Based on Real Data

Abstract

This paper describes a novel automotive snapshot integrity algorithm for bounding position, based on modelling GNSS measurements with non-Gaussian error distributions. A Bayesian method is used to derive the posterior probability distribution on position given a set of pseudorange and carrier phase observations from a single epoch. MCMC is then used to obtain rigorous probabilistic bounds on position. The MCMC method uses a novel form of parallel tempering to properly sample the multimodal posterior distribution created by carrier phase integer ambiguities, and importance sampling to obtain faster than real-time computational performance. Experimental results based on 27 hours of road driving show that integrity is maintained properly, with bounds which are significantly tighter than a more conventional EKF approach.