Research On Multipath Mitigation Algorithm Based On K-means And Sidereal Filtering In GNSS

The positioning and navigation services provided by the Global Navigation Satellite System(GNSS)have been widely used in various fields.In the city,due to the presence of a large number of high-rise buildings and reflectors,the reflection point of the multipath signal is closer to the target,which seriously interferes with the positioning accuracy.Because the multipath effect interferes with the positioning accuracy seriously,and its errors and parameters are difficult to accurately estimate through modeling,multipath has become the main error source in urban environments.In order to suppress the multipath effect,this paper studies the multipath suppression technique from two angles: multipath signal detection and inaccurate estimation of multipath repeat period,and the main research contents are as follows:(1)In view of the problem of multipath signal detection,it is proposed to use k-means clustering method to detect and classify multipath signals and direct signals,and the information needs to be extracted from the original data of observation when classifying GNSS signals,and the carrier-to-noise ratio,pseudo-distance,and satellite correlation values are mainly used in this paper to classify direct signals and multipaths.Then,the detected multipath is calculated by using a combination of least squares and autoregression to estimate the model coefficient of the code deviation,and a solution is used to construct a multipath delay to suppress the multipath to improve the positioning accuracy.Finally,experimental simulation shows that the k-means algorithm achieves a good effect in detecting and classifying multipath and direct signals.Experiments show that the recognition rate of the k-means algorithm for the direct signal is as high as 96%,and the recognition rate of the multipath signal is as high as 85%,compared with RBF and Liner,the classification accuracy is improved by 43% and 41%,respectively,and the multipath suppression algorithm used greatly improves the positioning accuracy.(2)In view of the problem that maneuvering of satellite orbits and long ephemeris update intervals will lead to inaccurate estimation of the recurrence period,it is proposed to use stellar filters to mitigate multipath errors.First of all,a simple and feasible real-time multipath filtering scheme is found to avoid the complex multipath repetitive cycle estimation process,and then the stellar filtering method based on sequence matching is used to effectively alleviate the multipath,taking the time series formed by the current almanac and the previous several epoche elements as the query sequence,and searching for matching sequences in the previous day’s time series.Similarity between sequences is assessed through similarity measures.Multipath suppression of the current almanac is performed by using the least squares method to determine the approximate linear relationship of the 2-day time series near the current almanac,and finally the stellar filtering method based on sequence matching can be effectively suppressed by experiments,and the multipath signal improvement after stellar filtering is about 27%.