Research On Outdoor Localization Algorithms Of LTE Terminal Based On Measurement Report

Network-side localization is one of the most important techniques for some wireless applications such as network optimization,emergency rescue and location services.In the long term evolution(LTE)system,measurement report(MR)is an important source of the user location information for network operators and content providers.However,the location information will not be included in MR when there is no global positioning system(GPS)module or the user does not cooperate to turn on the GPS on the mobile terminal.The MR-based network-side fingerprinting localization does not require user cooperation or additional hardware support,and instead of using GPS directly,it uses the other location-related information in the MR to locate the mobile terminals.In general,the fingerprinting localization process consists of an offline phase of fingerprinting database construction and an online phase of localization.During the offline phase,the MR with GPS information is collected at the network side and the fingerprinting database is constructed sequentially.During the online phase,the non-GPS-cooperation terminals are located in real-time using some fingerprinting matching algorithms.However,the large coverage area of fingerprinting database leads to a large searching space,which results in complexity issues during the online phase.In addition,the large fluctuation range of reference signal received power(RSRP)measurements and the small number of base stations that can be detected by the fingerprint are also the key factors to reduce the performance of the localization performance.This dissertation studies the key issues affecting the performance of the MR-based fingerprinting localization,the main work and innovation are as follows:Firstly,for the problem of the large searching space during online phase and the fluctuation of RSRP measurements,a new location method based on triplet fingerprint filtering is proposed,where the triplet fingerprint refers to Cell-ID,timing advance(TA)and base station environment.The fingerprints from the database which are far from the target terminal will be excluded by the filtering strategy before matching.In this way,the performance degradation caused by the fluctuation of RSRP can be made up to some extent,and the searching space can be reduced effectively.Then the range of TA at the location of the target terminal is analyzed to guide how to set the TA threshold in the filtering strategy.During the online phase,an improved matching algorithm WKNN is proposed to remove the abnormal fingerprints that are similar to the RSRP of the target terminal but far away from the target terminal.In this way,the localization performance can be improved further.At last,the proposed method is evaluated by MR dataset,experiment results show that the average location error of this method is 48 meters,and the computational complexity is reduced by 3?4 times after triplet fingerprint filtering.Secondly,for the problem of low localization accuracy when the number of detectable base stations is small,hidden markov model(HMM)is used to model the moving process of the target terminal in this dissertation,and MR sequence based fingerprinting localization algorithm is proposed.Due to the absence of sensor information,the network-side localization methods cannot calculate the transition probability of the target terminal through the direction and the distance between adjacent movements as the terminal-side do.In this dissertation,the transition probability is calculated based on the variation between the RSRP vectors at the current and previous moment to eliminate the need for any sensor.Then,we propose an adaptive switching strategy to suppress the cumulative errors by identifying and removing the fingerprints with large errors during the localization at the current moment.At last,the road network information is used to correct the localization results according to the typical moving characteristics of the target terminal.At the same time,the moving speed of the terminal is limited to ensure the continuity between the adjacent location results.The proposed method is evaluated by MR dataset,experiment results show that the proposed method can suppress the cumulative errors,and the average localization error is around 25 meters.