Indoor Localization Method Based On Location Fingerprint And Range Measurement

With the rapid spread of WiFi networks,WiFi-based indoor positioning technology has aroused wide attention for its low cost and easy installment.Among these technologies,WiFi-based passive fingerprint indoor positioning technology has gradually become the focus,because of the good non-invasive character.In general,the passive fingerprint positioning technology usually works in two phases: offline phase and online phase.The offline phase aims to construct the offline fingerprint database through the acquisition of the corresponding signal.While in the online phase,we need to match the online measured fingerprint with all the fingerprints in the offline fingerprint database,so as to estimate the target position.In the online phase,there are some fingerprints in the offline fingerprint database,which are far from the current position of the target,then they may interfere with the fingerprint matching,resulting in a big positioning error.For that reason,with the support of positioning technology based on location fingerprint and range measurement algorithm,this paper adds coarse positioning in the online phase.Before fingerprint matching,the coarse positioning filters the fingerprints that are not related to the current location of the target in the offline fingerprint database,avoiding the interference of irrelevant fingerprints and ensuring the accuracy of the positioning results.The main contents of this paper are as follows:(1)We propose an indoor localization method called ILLFRM(Indoor Localization Method Based on Location Fingerprint and Range Measurement).The method includes two parts: offline phase and online phase.In the offline phase,the offline fingerprint database is generated by using the generation method for fingerprints based on the principal component analysis(PCA).In the online phase,the offline fingerprint database is first filtered by the range measurement algorithm,and then we can determine the target location through the optimized fingerprint matching algorithm.(2)We establish the fingerprint database by using the channel state information(CSI)with better time stability and spatial distinguishability.Due to the high dimension of CSI signal,we propose a fingerprint generation method for fingerprints based on principal component analysis to reduce the computational complexity.Extracting the principal component from the CSI information through the PCA not only reduces the dimension of the CSI while preserving the main information associated with the location,but also eliminates some of the location-independent noise.Then,the weight of the principal component is set according to the corresponding variance of each principal component,so as to generate a fingerprint by using the principal component with weight.Experiments show that fingerprints generated at the same location of this method have better similarity than fingerprints generated by directly using CSI,while fingerprints generated at different locations of this method are easier to be distinguished.(3)We propose a CSI range measurement algorithm that is used for coarse positioning.Specifically,the range measurement algorithm is used to calculate the range between the target and the WiFi transceiver at first,and then using the range information to determine the area of the target location,that is,the method of coarse positioning.As a result,fingerprints that are not within the area can be removed from the fingerprint database.Compared with matching fingerprint directly in the online phase,coarse positioning can not only reduce the number of fingerprints needed to be matched,but also eliminate the interference of fingerprints outside the coarse positioning area.(4)In the fingerprint matching stage,the k-nearest neighbor algorithm is optimized by using the hierarchical clustering algorithm.And we applied the cluster analysis for the k reference points with the smallest Euclidean distance,and then chose one of the sub-clusters for the estimation of position.In order to achieve a better clustering effect,we take into account the signal characteristics and geographical location characteristics of the fingerprint of reference point.And with the help of geographical location characteristics,we correct the range of the fingerprint signals,acquiring a better positioning accuracy than that gained through the k-nearest neighbor algorithm.(5)We implement the indoor localization system based on location fingerprint and range measurement,and analyze as well as summarize the performance of the system by experiment.The experimental results show that the proposed method in this paper has a better accuracy for positioning than the existing WiFi-based passive fingerprint indoor localization method.