The Effect of Different Receiver Types and Ionospheric Conditions on Multi-GNSS Observable-Specific Biases

Chip shape distortions affect pseudorange signals of all GNSS satellites. In particular, analog components in the onboard signal generation chain lead to different distortions among the various satellites. Therefore, receivers with dissimilar correlator or front-end designs will yield incompatible biases in the pseudorange observable even for the same satellite. This inconsistency creates problems, when biases are estimated based on data from networks with mixed receiver types, like the IGS network, or if the user- and server-side receivers differ, like for example in open precise point positioning correction services. While this effect may be negligible in pseudorange-based single point positioning, convergence time and ambiguity resolution in high-precision applications will be adversely affected. Furthermore, constraints on the ionospheric parameters are required to make absolute or differential signal biases estimable. The choice of parametrization and constraints for the ionospheric delays together with the conditions of ionosphere will affect the estimation results. In particular, the day-to-day variability of daily bias estimates is notably increased during times of higher ionospheric activity. Different strategies for removing or estimating the ionospheric delays, for example a-priori corrections from GIMs, estimation of VTEC or STEC, will also have an impact on the results. The first part of the paper assesses the impact of different ionospheric conditions and parametrizations of the delay parameters on the estimation of biases. The goal is to identify a suitable strategy, which yields stable bias estimates in both calm and active ionosphere. Based on the selected parametrization, pseudorange biases are estimated for GPS, Galileo, BeiDou-3 and QZSS on multiple frequencies. Different sets of biases are estimated based on global networks of consistent receiver types. The results are then compared to illustrate the dependency on receiver type. This study uses the new approach of processing raw, uncombined observations for the estimation of observable-specific biases.