APPLICATION OF PCA AND KLE TO HIGHRATE POSITIONING
Ao Minsi 1) ; Hu Youjian 1) ; Zhao Bin 2,3) ; Ye Xianfeng 1) ; and Ding Kaihua 1)
1)Faculty of Information and Engineering, China University of Geosciences, Wuhan 430074; 2)Institute of Seismology, China Earthquake Administration, Wuhan 430071; 3)Research Center of GNSS, Wuhan University, Wuhan 430079
Abstract:Multipath error and random noise are two important error sources in highrate GPS positioning. In order to characterize and mitigate these errors, and improve the accuracy of highrate GPS positioning, the Principal Component Analysis (PCA) and Karhunenloeve Expansion (KLE) are introduced to evaluate the random noise level of the time series of coordinate on some days, extract and mitigate the multipath error from coordinates time series of multiple days. The data processing, comparison and analysis based on real data show that the principal component coefficients ratio of PCA can effectively reflect the random noise level of the time series of coordinates on certain day, meanwhile, the multipath error of the time series of coordinates on many days can be extracted and mitigated with introduction of PCA. Compared to PCA, the KLE approach performs slightly weaker on accurate improvement,but the random noise and local anomaly can be suppressed by KLE better since it is not sensitive to the random noise.