Statistical Anlysis Of GNSS Signal Characteristics For Complex Urban Environment

At present,the application of global navigation satellite system(GNSS)has become indispensable in the daily life,such as vehicle navigation,pedestrian positioning and so on.In the propagation space,GNSS signals are vulnerable to the interferences of surroundings,and the signal quality in different scenarios is quite different,which leads to the decline of positioning performance.Therefore,it is necessary to study the specific distribution of GNSS signal characteristics to assist the development and testing of receiver.In current applications,the GNSS signal simulator is usually used to adjust the characteristic parameters of the signal,so as to study the specific impact of signal interference characteristics on positioning accuracy.However,most of the signal feature models are based on the research of communication satellite signals in current simulators,which are inconsistent with the features of real GNSS signals,especially for the unique parameters in GNSS applications.In order to study the detailed characteristics of GNSS signals in practical applications,it is necessary to collect a large number of real satellite signals in various scenarios for analysis.Therefore,it is necessary to develop a multiscene real GNSS signal analysis technology,so as to obtain the actual feature distribution of navigation signals in various scenarios.The development of multi-scene real GNSS signal analysis system mainly includes two kinds of technical indicators,namely,the authenticity and comprehensiveness of the analysis results.To solve the problem of authenticity,it is necessary to ensure that the extracted signal characteristic parameters are accurate enough and can represent the corresponding scene types.The technical difficulty is to reduce the influence of nonenvironmental factors on signal parameters,and to classify a large number of collected signal data according to the type of scene correctly.To solve the problem of comprehensiveness,it is necessary to extract all the signal characteristic parameters that affect the positioning accuracy as far as possible,and analyze their statistical characteristics in detail.Multipath signal characteristics are one of the main factors affecting the positioning accuracy,and the research is relatively difficult.Therefore,in order to improve the authenticity and comprehensiveness of navigation signal analysis results,this paper mainly does the following four aspects:Firstly,a subsystem of real GNSS signal data collection and processing is established to extract signal characteristic parameters accurately,which provides data for the following study of signal characteristics model.In this paper,a GNSS software receiver is developed to execute signal acquisition,tracking,ephemeris calculation,positioning and other modules.A multipath signal estimation algorithm is added in the tracking module to achieve accurate characteristics extraction,such as signal strength,Doppler,multipath parameters,pseudorange,precision factor and satellite elevation.The test results show that the estimation error of signal power is less than 0.5d B;the estimation error of multipath delay is less than 10m;the estimation error of multipath signal power attenuation is less than 1d B;and the estimation error of multipath fading frequency is less than 1%.Secondly,a scene type recognition algorithm based on navigation signal characteristic parameters is proposed in this paper.It can be used to provide a reference for the algorithms switching of scene adaptive positioning algorithms.In this algorithm,signal power attenuation mean,signal power attenuation standard deviation,blocking coefficient,DOP growth coefficient and signal power fluctuation are used as eigenvectors for the recognition algorithm.Based on a large number of real signal sample data,scene types are recognized by using support vector machine classifier,and a timedomain smoothing method is established to improve the recognition accuracy.The experimental results show that the recognition accuracy of the algorithm is between 81.3% and 97.4% in each scene.Thirdly,the statistical distribution models of signal power attenuation and multipath signal parameters in urban canyon scene are studied,because the urban canyon is one of the most serious scenarios for signal interferences.For the characteristic parameters of signal power attenuation,a Mixture Gauss Distribution Model based on hidden Markov chain is established.For multipath signal parameters,the probability density distribution model of multipath delay and multipath fading frequency are established respectively,and the correlation between multipath signal strength and multipath delay is studied.In addition,the effects of satellite orbit and elevation on the parameters of each statistical model are further analyzed.Fourthly,the analysis shows that the multipath signal energy is high and signal occlusion is serious in the urban canyon scene,which makes it easy to produce NLOS signals.Therefore,the statistical characteristics of NLOS signals in this scenario are studied in depth,and a NLOS signal recognition algorithm based on signal characteristic parameters is proposed.The algorithm analyses the parameters such as signal intensity attenuation,pseudorange carrier difference rate and pseudorange residual,and establishes the posterior probability distribution of the parameters.So it can detect the NLOS signals in the received signal step by step through multiple iterations,and improves the positioning accuracy.In conclusion,this paper focuses on the multi-scene real GNSS signal analysis technology.Aiming at the problems of feature parameter extraction,scene type classification and statistical analysis of GNSS signal,the statistical distribution of GNSS signal characteristics in complex urban environment is deeply studied.The results provide effective technical support for the development and testing of GNSS receiver.