Observation prediction strategy of GNSS RTK reference station based on first-order model compensation

In this paper, high-precision method to predict observations from reference station of real-time kinematic (RTK) is proposed, which may be caused by the interruption of reference station communication. The satellite ephemeris, ionosphere, troposphere and other error sources that affect the prediction of the Global Narigation Satellicte System (GNSS) observations are analyzed, and the sensitivity of various error sources to the length of delay is compared. Results show that the forecast error accumulates linearly as the delay increases, and is negatively correlated with the satellite elevation. When the elevation angle is lower than 10°, the error caused by the precise ephemeris forecast for 5 minutes can reach 174.6 cm. At the same time, the linear trend can be calculated when the reference station data is not missing which can be used to compensate predicted observations. Under the same conditions, cumulative error of compensated observation is reduced to 64.4 cm. The RTK positioning results show that when delay is 1 min, the RMS of the three directions of E, N, and U are 0.37 cm, 0.41 cm, and 0.86 cm respectively, which is 71.1%, 77.2% and 90.0% better than that without compensation. By comparison with 5 minutes’ delay, the centimeter-level accuracy can still be maintained.