Ionosphere Augmentation for Accelerated Convergence in Precise Point Positioning with Triple-frequency and GPS

Abstract

A Precise Point Positioning (PPP) method is presented that uses raw triple-frequency GPS measurements augmented with a regional ionospheric model (RIM) for faster solution convergence. The proposed method has two characteristics. Firstly, there is no noise amplification since raw uncombined carrier phase and code measurements are used. This reduces the solution convergence time, contrary to the ionosphere-free linear combination with a noise amplification factor of 3 that is used in traditional dual-frequency model. Secondly, the ionosphere delay remaining in the uncombined equations is modelled using externally provided precise ionosphere corrections, hence reducing the number of parameters to be estimated. Triple frequency simulated data was used to validate the model using a fully operational GPS constellation at four sites in Australia for one week duration. The performance of 2-hourly PPP solutions were compared for three cases including (1) the proposed triple frequency raw uncombined model with treatment of the ionosphere delay using an externally provided model (2) triple frequency raw uncombined model with estimation of ionosphere error (i.e. without RIM) and (3) traditional dual frequency PPP model. The three cases were compared in terms of solution convergence time taken to achieve and maintain 3- dimensional (3D), horizontal and vertical positional accuracies of 1 decimeter and 5cm. The tests for the first case considered the use of varying precisions of the externally provided ionosphere model with standard deviations (std) of 0.1, 0.25, 0.5 and 1.0 TECU. Results from the comparison show that the best improvement was achieved with the proposed raw model when using a high precision ionosphere model with std = 0.1 TECU, where the convergence time to achieve sub-decimeter accuracy was reduced by 37%, 84% and 33% in the 3D, horizontal and vertical components. The most significant improvement was in the horizontal component since the ionospheric error mostly affects the height. The second model also outperformed the standard dual frequency model, but had less improvement in convergence time by 13%, 3% and 21% in the three components.