Outlier Detection Performance in Precise Point Positioning

With the development of iono-free solution, provision of IGS orbit and clock products, high precision GNSS positioning with a single receiver, usually named Precise Point Positioning (PPP), has been playing an increasingly important role in many applications. PPP is to obtain the highly precise coordinates of the receivers, so it is necessary to establish the mathematical model according to the mechanism of the motion of the platform and geometric relationship for the measurements used. The commonly used method to describe these motion and relationship is the Gauss-Markov models which are then used in the Kalman filter, to produce optimal estimation of the parameters, only when the model is realistic. However, it is well known that GPS/GNSS measurements may be contaminated with outliers, which will certainly have an impact on the reliability of the results, and even the convergence period. If these outliers are not correctly identified, the results may be biased and the precision will be decreased. Therefore, it is necessary to analyse the outlier detection performance first. The experiment demonstrates that the internal reliability gives the minimal detectable bias of 4cm in carrier phases and 3.5m in pseudo-ranges under the given priori variance and the effects of undetectable bias on the states is at the millimetre level after convergence.