Research Of Indoor Fingerprint Location Based On Channel State Information

Indoor localization technology is the core technology based on location information service technology.Up to now,passive indoor localization technology based on channel state information(CSI)is one of the research hotspots in positioning technology.Channel state information is a channel attribute on a data transmission link in the field of wireless communication.Compared with the traditional location based on signal reception strength(RSS),CSI is more convenient to indicate different locations in space.CSI is characterized by a large number of signals,sensitivity to millimeter changes,frequency diversity and spatial diversity,etc.Because wireless LANs are everywhere in life.The use of WiFi signals for indoor localization research on CSI has become an important topic in the field of channel state information applications.In the existing CSI-based localization methods,the localization method based on multi-location estimation fusion is still relatively blank.Based on the in-depth study of indoor fingerprint localization based on CSI,this thesis combines with the deep learning algorithm in recent years to carry out multi-model.The main content of the indoor localization research is as follows:(1)Based on the sufficient analysis of the channel state information data collected during the time period,it is verified by experiments that the CSI has frequency diversity and spatial diversity,which provides a theoretical basis for the subsequent positioning system.Aiming at the problem of abnormal data during the acquisition process,it is proposed to use the anomaly detection algorithm(LOF)to process the abnormal data in the data set to ensure the accuracy of the classification model.(2)Based on the in-depth study of deep learning algorithm and group method of data handing(GMDH),this thesis proposes a new passive indoor positioning algorithm(GMNN),which transforms the continuous spatial localization regression problem into discrete localization classification problem.Fully collect CSI data in space,obtain multiple positioning results,and use GMDH algorithm to combine multiple positioning estimates to obtain the final positioning target estimation.This thesis gives the specific practice of the GMNN system.The comparison experiments show that the indoor localization method proposed in this thesis has a certain degree of improvement compared with the same type of localization method,and it can effectively avoid the loss of positioning accuracy of some equipment.The problem of increased overall system localization error.