Network-Based Stochastic Model for Instantaneous GNSS Real-Time Kinematic Positioning

Abstract

The concept of global navigation satellite system (GNSS) real-time kinematic (RTK) positioning through the use of multiple reference stations (Network RTK) is the most common approach to relative positioning, which makes it possible to achieve centimeter-level accuracy for medium baselines. In this approach, ionospheric and geometric correction terms, generated on the basis of a model of interpolation of the distance-dependent biases, are applied to the functional model of rover positioning. The accuracy and reliability of Network RTK performance depend on the accuracy of the defined correction terms. Especially during storm-level ionospheric activity, the applied spatial interpolation model might not be suitable for the real ionospheric state, causing the ambiguity resolution to be less reliable, or even impossible, because of high residual errors. Thus, the residual errors can substantially degrade the correctness of the functional model and should be accounted for to obtain optimal estimation of the unknowns in the positioning model. One of the possible approaches for taking into account such errors is to introduce them into a stochastic model rather than a functional model. This paper provides a method of taking into account residual errors in the stochastic description of the positioning model by using the accuracy characteristics of the correction terms directly defined in the network solution. It describes a method of developing the proposed stochastic model (called the Network-Based Stochastic Model), including of the test results of the instantaneous Network RTK positioning performance.