RTK-quality Positioning with Global PPP Corrections

Abstract

Global Navigation Satellite System (GNSS) Precise Point Positioning (PPP) processing has been subject to much research in recent years, and the technique has shown potential as an alternative or replacement for Real-Time Kinematic (RTK) processing. Indeed, RTK (and network RTK) is known for being able to reach cm-level positions instantaneously, or within a few seconds, primarily depending on the length of the baseline. However, local reference stations are required for measurement differencing to eliminate error sources, which is a limitation in areas where such infrastructure is not available or costly. PPP on the other hand does not need local stations, and only requires global satellite corrections to achieve high precision positions. This work reaches for RTK levels of performance without the need for local information or infrastructure through PPP. Conventional PPP is a measurement-dependent technique requiring tens of minutes to reach centimetre-level positions. Luckily, given the recent advances in GNSS with many constellations broadcasting on two, three, or four frequencies, this work makes use of the signals currently available by processing signals from GPS, Galileo, BeiDou-2/3, and GLONASS and fixing ambiguities for the first four constellations, on as many frequencies as available. This processing is done through a four-frequency, five-constellation uncombined Decoupled Clock Model (DCM) that has been expanded as part of this work, and the results are tested on 1448 datasets from stations spread around the world. Results show that sub-decimetre positioning can be achieved instantaneously at 2? and 1?. Compared to RTK, the proposed algorithm is able to reach instantaneous convergence of 2.5 cm for 81% of stations, and within one minute for 89% of stations, showing that PPP can indeed consistently reach centimetre-level positions quasi-instantaneously. These results are reinforced when performing epoch-by-epoch processing without use of a priori information, as an average of approximately 80% of all epochs from 27 stations processed over one week converge below 2.5 cm in a single epoch at 1?, as opposed to less than 0.5% for a classic PPP solution. These results demonstrate that consistent RTK-like quality can be achieved with PPP processing without the need for local infrastructure or corrections, opening the door for truly global precise positioning.