Research On LTE Region Localization Algorithm Based On Distributed Antennas

With the rapid growth of mobile devices,the demand for location-based services is increasing rapidly.Nowadays,Global Positioning System(GPS)can provide highprecision positioning service for outdoor environment.However,because of the serious obstruction of walls and other obstacles,indoor positioning technology has been developing slowly.The emergence of Distributed Antenna System(DAS)effectively solves the problem of indoor blind spot coverage of communication signals.Hence DAS based indoor localization technology has become a research hotspot.In particular,DAS uses antennas scattered in indoor to send operator signals and the antenna may malfunction during operation.Since all antennas of the DAS have the same Mac address,there is no way to know the location of the faulty antenna.Therefore,it is necessary to locate the users in the area to find the fault antenna.Based on this,this paper has carried out the LTE region localization algorithm for distributed antennas.The research contents are as follows:1.In order to save the construction time cost and achieve reasonable regional division,the improved fast regional database construction method is studied.The realtime velocity and heading estimation are carried out by gait detection and extended Kalman filtering of arbitrary motion sequences,and position solution and signal mapping are completed.Next,signal features of sub-regions are transformed into distance ratio features of unit categories through data combination,and similarity clustering is carried out according to these features.Finally,coordinate distance is used as a limiting parameter to update iteratively,so as to ensure the continuity of data points.2.In order to characterize the similarity of signals between sub-regions and improve the matching probability,the improved fast Earth Mover's Distance(EMD)algorithm is studied.Firstly,according to the difference of probability distribution of regional signals,different distribution distance models are constructed and converted.Secondly,the restricted parameters within and between regions are calculated to construct special unit signal distance.Finally,the unit signal distance is continuously extracted and updated to achieve fast EMD distance calculation.3.In order to improve the accuracy of regional location under single station identity and low sampling rate signals,the regional location algorithm based on EMD and Support Vector Machine(SVM)combined classification is proposed.Firstly,considering the difference of sampling rate between databases and measured data,the feature of adjacent signals is extracted by recursive search of reference location points.Secondly,the joint classification model of EMD and SVM is constructed by the correlation between EMD distance and SVM classification.Finally,the regional location is achieved by EMD matching and comprehensive voting of joint classification results.The experimental results show that the proposed algorithm saves a lot of database construction time in the offline phase.The accuracy of regional positioning reaches over 90% under dynamic test data,and the static test data reaches more than 70%.