Study On Short Baseline Carrier Phase Multipath Error Modeling Based On Time-Space Method

In the Global Navigation Satellite System(GNSS)relative positioning application,the multipath error isn't negligible.For short and medium baselines,most errors,such as tropospheric delay,ionospheric delay,satellite orbit error,and satellite clock error,can be eliminated or minimized by differential technology,but the multipath error is difficult to eliminate by this method.Therefore,in high-precision positioning applications,the multipath error is always regarded as the main source of error.Though some hardware-based technologies can be utilized to mitigate most of the code multipath effects,carrier phase multipath still exists in most cases.In static or quasistatic scenarios,the multipath error is systematic to some extent,rather than purely random,and should be treated as a signal rather than noise.Based on this fact,it can be modeled empirically by using repetitive features.In most cases,the modeling process can be carried out in the time domain or the space domain.The models obtained by the two methods are functions of time and satellite position,respectively.Aiming at the above research problems,this paper has studied the characteristics of the multipath error,common processing and correction techniques,time domain modeling methods of the multipath error,space domain modeling methods and so on.The main work and results of this paper can be summarized as follows.1.We have deduced the characteristic of the multipath error.The mathematical model of the multipath error is derived and its causes are analyzed.Aiming at several main factors that affect the multipath error,that is,the reflection coefficient of the reflector,the distance between the reflector and the receiver antenna and the incident angle of the reflected signal,experiments under different conditions are designed to observe and analyze the influence on the multipath error.2.A new time domain modeling denoising method is studied and proposed.First of all,according to the smoothness of the multipath signal,the L1 regularization method is introduced to filter out the random noise in the residuals series,so as to extract the pure multipath signal to establish the multipath error model.Then,how to determine the repeat period of the multipath error in time domain modeling is studied.The multipath repeat time(MRT)of GPS satellites is not exactly equal to sidereal days,and accurate MRT helps to use sidereal day filtering.Three methods: Orbit Repeat Time Method(ORTM),Aspect Repeat Time Adjustment(ARTA),and Residuals Correlation Method(RCM)are used to estimate the MRT,and the respective effects are compared and analyzed to prepare for the subsequent time domain filtering model establishment.Secondly,aiming at the limitation of time domain method on data,piecewise cubic spline interpolation and piecewise linear interpolation are introduced to establish a continuous sidereal filtering model.Finally,the multipath error denoising effect of the L1 regularization method proposed in this paper is compared with that of the traditional wavelet method.3.A new space domain modeling and denoising method is studied and proposed.Drawing on the excellent results in the field of machine learning.The k-means clusterassisted RBF neural network learning method is used for space domain modeling,simulation experiments with simulated noise are used to verify the noise reduction effect of the model,and measured data are used to further verify the training the ability of noise reduction and the multipath model estimation.In the coordinate domain,it is compared with the advanced sidereal day filtering method to verify the effectiveness of the proposed space domain method.The paper has 39 pictures,19 tables,and 116 references.