Application of PCA and KLE in high-rate GPS positioning

摘要

Multipath error and random noise are two important error sources in high-rate GPS positioning. Some methods, including those based on sidereal day filtering and digital signal processing, have been developed to mitigate multipath errors. To improve the accuracy of high-rate GPS positioning, this paper uses Principal Component Analysis (PCA) and Karhunen-loeve Expansion (KLE) to evaluate the random noise level and mitigate the multipath error in the coordinate time series of GPS stations. The experiment results show that the response ratio of the principal component coefficients of PCA can effectively reflect the impact of strong random noises on coordinate time series, and the multipath error can be significantly mitigated by PCA transformation. Compared to PCA, the response ratio of the principal component coefficients of KLE is less sensitive to random noise and its ability to reduce multipath error is also weaker. However, KLE can preferably suppress random noises and local anomalies.