Modelling of dispersive and non-dispersive effects on network-based positioning

Abstract

The concept of network-based positioning has been extensively developed in order to better model the distance-dependent errors contained within the GPS carrier-phase measurements. These errors can be separated into a frequency-dependent or dispersive term (e.g. the ionospheric delay) and a non-dispersive term such as the tropospheric delay and orbit biases. Dispersive and non-dispersive errors have different dynamic effects on the linearity of the network correction. Dispersive errors change rapidly and have strong variations, therefore the network correction has to be generated as frequently as possible (typically epoch-by-epoch) to mitigate their effect. On the other hand, non-dispersive errors change smoothly over time, and a similar attempt to model these errors will have a tendency of increasing the residuals. For this reason it is suggested that the non-dispersive term should not be modelled on an epoch-by-epoch basis. In this study, a simple running average is proposed to provide a stable network correction for the non-dispersive term. It is found that the correction can be used to obtain better ionosphere-free measurements, and therefore helpful in resolving the integer ambiguity of the original GPS carrier-phase measurements. Once the integer ambiguities have been resolved, dispersive and non-dispersive corrections can be applied to the fixed carrier-phase measurements for the positioning process to improve the accuracy of the estimated coordinates.