Receiver Bias Estimation Strategy in the Uncombined Triple-Frequency PPP-AR Model

Abstract

This study investigates the reparameterization of the uncombined triple-frequency PPP-AR model, mainly in terms of the receiver hardware bias estimation. We explore the impact of the number of estimated receiver bias parameters as a function of pseudorange noise, i.e., the trade-off between estimating too many bias parameters on cost of a high stochastic error posing a challenge on ambiguity resolution on one hand, and estimating too few bias parameters on cost of ignored inconsistencies on the other hand. We implemented 4 different bias estimation strategies and compared their performance in positioning and ambiguity resolution against each other in the presence of phase bias across various pseudorange noise levels. The results show that with accurately initialized reference ambiguities, for code noise levels below 0.3 meters, estimating four biases (one each for P3, L1, L2, L3 signals) outperforms other strategies, while for code noise levels exceeding 0.3 meters, estimating two biases is sufficient. Conversely, with inaccurately estimated reference ambiguities, estimating four biases constantly prevails across all code noise levels. In ideal conditions, i.e., bias-free scenario, however, estimating only one bias is the optimal choice. This research enables readers to get insight into bias estimation strategies in the uncombined triple-frequency PPP-AR model and their impact on positioning performance and ambiguity resolution across different code noise levels. The conclusions can act as a guideline supporting the user implementation of the optimum representation of hardware biases in the uncombined PPP-AR model.