Font Size: a A A

Multi-dimensional Information Fusion Localization Algorithm Based On 5G Technology

Posted on:2022-11-23Degree:MasterType:Thesis
Country:ChinaCandidate:W Q JiFull Text:PDF
GTID:2518306773985329Subject:Automation Technology
Abstract/Summary:PDF Full Text Request
5G is the latest technology in mobile communication technologies so far.Users will generate a lot of data when they are surfing the web.Researching that data for user's location prediction plays a central role in the design of 5G communications and lo T wireless technologies.Based on 5G communication technology,this paper constructed a multi-dimensional information fusion fingerprint positioning method.And conducting the following studies on the construction and optimization of fingerprint database in indoor and outdoor environments respectively.1)In view of the phenomenon that the feature discrimination of fingerprints is inconspicuous when using RSSI to construct fingerprint database for positioning in traditional method,which lead to a poor position accuracy.Multi-dimensional information fusion fingerprint algorithm based on 5G is proposed.The fingerprint database is constructed by using a distance-related power signal,which is combined with the MLP algorithm for positioning.Experiments have verified that compared with the RSSI fingerprint,the position accuracy of this method has been improved by 19.58%,10.70% and 12.98% in three different indoor experimental scenarios.And the position accuracy has reached 3m requirement mentioned in 3GPP Rel-16.The best position accuracy is 1.594 m when CDF is 80%.2)To solve the problems of correlation and data redundancy among variables in the fingerprint database of multidimensional information fusion,A multi-dimensional information fusion optimization algorithm combined with Random Forest Variable Selection was proposed.The goal is to optimize the multi-dimensional information fusion fingerprint algorithm based on5 G proposed in this paper.By calculating the importance of each variable and setting the threshold,the best fingerprint combination is selected to construct the RFVS fingerprint database and positioning.Results indicate that RFVS fingerprint database improves the accuracy of 39.56%,30.54%,30.84% compared with the RSSI positioning in three experimental scenarios.Comparing with multi-dimensional information fusion fingerprint,the improvement rate is 24.84%,22.21% and 20.52%.And the best position accuracy is 1.198 m when CDF is 80%.3)In view of the difference between indoor and outdoor base station antenna configurations and observed quantities,a method based on multi-beam information fusion fingerprint database is raised based on 5G Massive MIMO technology.We conducted the experiment outdoors,and researched the influence of sampling method of test points and the amount of adjacent base stations.Results indicate that the improved rate of multi-beam information fusion fingerprint positioning is more than 30% compared with single beam.The location accuracy of random sampling is better.The best performance was achieved by combining two adjacent base stations,and the improvement rate was 23.62%.The position accuracy reaches the 20 m requirement outdoor which is mentioned in 3GPP REL-16,and the best positioning accuracy reaches12.774 m when CDF is 80%.
Keywords/Search Tags:fingerprint positioning, 5G, multi-dimensional information fusion, random forest variable selection, multi-beam information fusion
PDF Full Text Request
Related items