Multipath Mitigation For BDS GEO Satellites Based On SVR

Multipath error is one of the error sources that badly affects the quality of GNSS services,and can't be ignored especially in high accuracy applications.And it even has become a bottleneck of the improvement of the quality of GNSS services,because there are still no very good solutions to it due to its close relation with the changeable environment.Besides that,studies have shown that the multipath error of BDS GEO satellites is different from that of MEO satellites,which may be due to the difference between their orbits.the multipath error of BDS GEO satellites can be described as low frequency component and high frequency component.This makes the performance of multipath mitigation of GEO satellites not so good or even ineffective by using traditional methods.This paper analyzed the data characteristics and the physical mechanism of the multipath GEO satellites,then lead to the SVR model of the multipath GEO satellites.The main contributions are as follows,Firstly,we analyzed the complexity of the multipath error of BDS GEO satellites in detail in time domain,frequency domain and time-frequency combination domain.And the made a brief analysis of the data quality of some iGMAS observation station.Secondly,we detailedly analyzed the physical mechanism of the complexity of the multipath error of BDS GEO satellite,then derived the mathematical formula that can explain that complexity,which is mainly caused by the rate of change of satellites elevation and azimuth.And it differs in different satellites for a specific observation station.At last,the SVM(Support Vector Machine)is the first to be applied in building the regression model of code multipath of GEO satellites in this paper.Then we come to a conclusion of that elevation and the rate of change of elevation are the best combination for building the SVR model,of which the performance is different among different satellites or different frequencies.Generally,the performance of the model of B2 frequency and B3 frequency is better than that of B1 frequency.And the best result that we have reached is 70.1% mitigation for B2 frequency.