Comparison and Development of Ionospheric Correction Methods in GNSS

Abstract

The ionosphere is one of the largest error sources in GNSS-based positioning, causing a delay in the arrival of the navigation signal. There are some existing correction methods, namely the Klobuchar model, the NeQuick G model and SBASs. None of these methods is simultaneously accurate, computationally light and having a low requirement of amount of data transfers. In this work the available ionospheric correction methods are studied and a new data-driven approach to obtain the corrections is developed. Two different sets of basis functions, polynomial and trigonometric, are tested for parametrization of the vertical ionospheric delay values transmitted by EGNOS. Universal kriging is applied on vertical ionospheric delay values obtained from dual frequency measurements. Furthermore, the parametrization of the vertical ionospheric delay values is predicted using a standard Kalman filter. The Kalman filter is tested using both a standard random walk state model, and a Klobuchar-based state model, which was developed during this work. The tests show that a compact presentation of the ionospheric delay with a small degradation of the resulting estimate is possible. Universal kriging proved to be challenging, but the obtained results still look promising in regards of the future work. Most essentially, the tests show that the parametrization and prediction of the vertical ionospheric delay outperform both the Klobuchar model and the NeQuick G model, and reduce the amount of data transfers needed significantly in comparison to the EGNOS transmissions.