Improved convergence for GNSS precise point positioning

Abstract

The precise point positioning (PPP) methodology allows for cm-level positioning accuracies using a single GNSS receiver, through careful modelling of all error sources affecting the signals. Adoption of PPP in several applications is however muted due to the time required for solutions to converge or re-converge to their expected accuracy, which regularly exceeds 30 minutes for a moving receiver. In an attempt at solving the convergence issues associated with PPP, three aspects were investigated. First, signal tracking interruptions are typically associated with integer discontinuities in carrier-phase measurements, often referred to as a cycle slips. A refined method for detecting and correcting cycle slips was thus developed, in which all error sources affecting the observations are either modelled or estimated. Application of this technique allows for instantaneous cycle-slip correction, meaning that continuous PPP solutions can be obtained even in the presence of short losses of lock on satellites. Second, external information on the ionosphere allows for reduced convergence times, but consistency must be observed in the functional model. A new technique, termed integer levelling, was thus developed to generate ionospheric delay corrections compatible with PPP based on the decoupled-clock model. Depending on the inter-station distances in the network providing ionospheric corrections, instantaneous cm-level accuracies can be obtained in PPP. Third, processing of GLONASS signals is more problematic than GPS due to frequency division multiple access, leading to inter-frequency carrier-phase and code biases. A novel approach for the estimation of such biases was then proposed and facilitates processing of mixed receiver types. It also allows for undifferenced GLONASS ambiguity resolution based on a heterogeneous network of stations, the first demonstration of such an approach ,and therefore has the potential to further reduce PPP convergence times. This research also emphasized potential benefits of integer-levelled observations for improved ionosphere monitoring. The main justifications for adopting this approach are: a reduction in the determination of slant total electron content errors, a simplification in the GLONASS processing strategy, its applicability in real time, and the generation of satellite biases required for the use of ionospheric constraints in PPP with ambiguity resolution.