The use of ambiguity resolution for continuous real-time frequency transfer by filtering GNSS carrier phase observations

We have studied the influence of phase ambiguities and their resolution on the quality of the station clock parameter when filtering real time GNSS carrier phase observations. Especially the benefits of filed integer phase parameters are discussed in this paper. An earlier work [1] describes a continuous real time GPS carrier phase processing software based on Kalman filtering of common view satellite observations. In this processing carrier phase ambiguities were handled by estimating separate initial L3 phase parameters for each observed satellite. In such an approach the frequency estimation suffers from mapping of differential station coordinate errors into these non integer start phase estimates. Most of the phase parameters are initialized with a biased error due to the fact that the orientation of satellite passes is unequally distributed, which in turn accumulates into the estimated station clock parameters. This introduces a constant error in the frequency estimation of the station clocks. Filed integer phase parameters are one solution to the problem described. The paper describes how to resolve integer phase ambiguities on single differences from a real time data stream. At startup, at any time a new satellite is included to the computation and at any cycle slip occasion, the differential phase ambiguities are resolved ahead of the filtering process. In such a way the initial phase parameters do not need to be estimated by the filter and leaks of position errors into the clock parameters are minimized, i.e. they do not bias the estimated frequency. Aditionally, the Kalman filter is less computational demanding, which allows higher throughput and more stations to be computed at the same time. Even though the approach was only tested on GPS carrier phase observables, the method is generally applicable for any multi frequency GNSS carrier signal.