Network real-time kinematic data screening by means of multivariate statistical analysis

We introduce a novel approach to the computation of network real-time kinematic (NRTK) data integrity, which can be used to improve the position accuracy for a rover receiver in the field. Our approach is based on multivariate statistical analysis and stochastic generalized linear model (SGLM). The new approach has an important objective of alarming GNSS network RTK carrier-phase users in case of an error by introducing a multi-layered approach. The network average error corrections and the corresponding variance fields are computed from the data, while the squared Mahalanobis distance (SMD) and Mahalanobis depth (MD) are used as test statistics to detect and remove data from satellites that supply inaccurate data. The variance-covariance matrices are also inspected and monitored to avoid the Heywood effect, i.e. negative variance generated by the processing filters. The quality checks were carried out at both the system and user levels in order to reduce the impact of extreme events on the rover position estimates. The SGLM is used to predict the user carrier-phase and code error statistics. Finally, we present analyses of real-world data sets to establish the practical viability of the proposed methods.