Protection Level for Precise Point Positioning Based on Gaussian Mixture Model

In the integrity assessment for Precise Point Positioning (PPP), the Protection Level (PL) is usually calculated based on the assumption of Gaussian white noise. However, the code and phase noise don’t follow the Gaussian distributions due to the multipath effect in some situations. For this reason, it makes the PL too conservative by simply assuming the noise follows a Gaussian distribution. To deal with this problem, a PL calculation algorithm for PPP based on Gaussian Mixture Extended Kalman Filter (GMEKF) is proposed in this paper. Firstly, Gaussian Mixture Model (GMM) is introduced to accurately describe the non-Gaussian characteristics of the noise. Secondly, the sub-filters and corresponding test statistics are constructed for each independent Gaussian component. Finally, the PL calculation principle is deduced based on the concept of integrity risk. The results show that the GMM can better de-scribe the non-Gaussian feature of the observation noise. In the suburban scenario, the HPLs and VPLs based on GMEKF are reduced by 33.6% and 33.1% compared with Kalman filter respectively, so as to improve the availability of the PPP integrity monitoring algorithm.