Research On Key Technologies Of Fusion Indoor Localization Based On UWB And PDR

With the development of wireless localization technology,needs of applications based on location information has been greatly improved.Indoor localization can provide various and convenient services for a variety of indoor activity scenes,which is an important research hotspot in the field of localization.However,the increasing needs for localization in complex indoor scenes poses a greater challenge to indoor localization methods.As a rising wireless transmission technology,ultra wideband(UWB)technology,with advantages such as the resistance of signal fading and multipath,occupies a place in the field of indoor localization,and tends to integrate with mobile devices,so as to provide more convenient localization services.This paper starts with the research of effective real-time localization scheme in typical indoor complex scene of city,analyzes the application feasibility of indoor localization based on UWB,and realizes a fusion localization method of UWB and PDR by integrating the PDR algorithm in inertial navigation system localization,so as to make the advantages of localization technology complementary to overcome the shortcomings of single localization method.In this paper,the main problems and key technologies are studied to realize a more suitable fusion localization method for complex scenes.The main research contents are as follows:(1)This paper analyzes the requirements and characteristics of localization application in complex indoor scene,evaluates the feasibility,advantages and key problems of UWB Technology in complex scene localization from the perspective of system and algorithm,analyzes the sources of errors that lead to the reduction of localization accuracy.(2)In order to solve the NLOS propagation problem in UWB localization,this paper introduces machine learning theory to improve the NLOS discrimination method,and transforms it into a binary classification problem.In view of the changes of signal characteristics caused by NLOS propagation compared with LOS propagation,UWB signal and channel characteristics are extracted.The feature subset is selected by feature subset selection strategy as the discrimination basis,and Support Vector Machine(SVM)is used to achieve the discrimination effect that can be completed without prior knowledge.Meanwhile,the high complexity and poor accuracy of traditional discrimination methods are overcome.(3)Aiming at the influence of NLOS propagation on UWB localization accuracy,this paper improves the acquisition method of PDR algorithm initial position information when NLOS propagation,introduces Generalized Extension Approximation model to process UWB localization data,makes up for the lack of localization information when NLOS propagation occurs,gives the best estimation of localization data,and provides more accurate PDR initial position reference.(4)In this paper,particle filter algorithm is used in the process of multi-source fusion localization which is more suitable for complex scenes.The contradiction between particle degradation suppression and particle diversity retention in the process of particle filter implementation is analyzed.The resampling method of particle is compared and analyzed.The experimental verification of the fusion system is carried out for the effective resampling method.The location error evaluation verifies that the method in this paper can overcome NLOS propagation and achieve better location performance,and meet the higher demand of location-based information research and service.