Enhancing multi-station URTK over long baselines using BDS PPP-B2b and Galileo HAS real-time precise corrections
Abstract
Long baselines and discontinuous communication links present significant challenges for real-time high-precision positioning. In sparse reference station networks, conventional Network Real-Time Kinematic (NRTK) suffers from amplified long-baseline errors and relies on continuous communication links. Precise Point Positioning (PPP) offers global applicability but suffers from slow convergence due to the absence of external atmospheric constraints. Undifferenced Network RTK (URTK) maintains a unified network datum and provides an absolute-positioning model consistent with PPP, without relying on a fixed main station in real-time processing. Building on these advantages, this study develops a real-time positioning framework that integrates URTK with B2b/High Accuracy Service (HAS) precise corrections for long baselines and asynchronous conditions. The computation platform applies B2b/HAS orbit and clock corrections to mitigate satellite-related errors at both platform and terminal. Fixed double-difference ambiguities are mapped to undifferenced integers to generate undifferenced corrections. Atmospheric delays and residual orbit errors are interpolated with multi-station regional models, forming comprehensive URTK corrections that contain integer ambiguities, fractional-cycle biases, and atmospheric terms. The terminal removes these components and achieves rapid ambiguity resolution with centimeter-level accuracy. It further employs an ionospheric weighted model driven by correction age to distinguish interpolated ionospheric residuals from asynchronous variations, thereby maintaining high-accuracy PPP-form positioning during correction interruption. Experimental results show that the proposed method achieves horizontal and vertical accuracies of 1.30 and 5.67 cm, respectively, in the multi-station case, representing the maximum improvement of 17.8% over the broadcast-ephemeris solution. Under simulated correction interruptions, it maintains fixed ambiguity solutions for over 20 min, whereas the broadcast-ephemeris solution degrades to the decimeter level positioning accuracy.