New Insights of Accurate Multipath Mitigation for Multi-GNSS PPP Using Refined Multipath Extraction Strategy

Multipath continues to pose a significant challenge for the Global Navigation Satellite System (GNSS) technique in achieving precise Precise Point Positioning (PPP). Sidereal Filtering (SF) and multipath hemispherical map (MHM) represent common methods for mitigating multipath by leveraging satellite temporal and spatial repeatability. However, the effectiveness of these approaches relies heavily on the quality of the multipath correction model derived from preceding PPP residuals, which often contains product and parameter estimation errors. These errors can notably compromise the accuracy and reliability of multipath mitigation efforts. We propose an innovative method for extracting multipath based on a refined error separation strategy and mitigating multipath with SF. This method involves decomposing PPP residuals into multiple reconstructed components (RCs) using the multi-channel singular spectrum analysis (MSSA) technique and subsequently isolating multipath by reconstructing RCs exhibiting strong temporal repeatability. Extensive experiments with a 28-day dataset from 20 multi-GNSS stations demonstrated the effectiveness of our method. Compared to the conventional wavelet-based SF method, the proposed approach improved PPP accuracy by 23%, 20%, and 18% in the East, North, and Up directions, respectively, and by 18%, 16%, and 15% compared to the MHM method. Results also reveal that PPP residuals can be decomposed into multipath, high-frequency noise, common-mode error (CME), and site-specific errors. The last two components, which are non-repeatable in time and space, pose limitations on the effectiveness of conventional SF and MHM strategies in addressing multipath effects.