Research On Indoor Positioning And Tracking Technologies Of ZigBee Based On Location Fingerprint

With the development of mobile internet technology,the higher demand for accurate location services is needed.In the outdoor,four major satellite systems,such as GPS and Beidou,can basically satisfy the requirements.However,the rapidly decreased intensity and quality of satellite signal leads to the "last mile" problem of indoor location services in the indoor.At present,a platform with ZigBee communication network to realize the location fingerprinting is a reasonable solution to the above mentioned problem.This paper,as the research target that providing accurate and real-time location informantion for indoor user,studies several key links of location fingerprint-based indoor positioning and tracking technologies using ZigBee.Firstly,for improving the reliability,method combining the theory with practice is used to analyze the characteristics of received signal strength.Aim at the time-varying characteristic of received signal strength,variance filter is adopted to pretreat the sample received at the offline stage.The experiment results manifest that distortion values are effectively eliminated and the radio map is constructed by this pretreatment,which improves the positioning accuracy.Secondly,In order to improve the matching efficiency of location fingerprint,the classic clustering algorithm k-means is introduced to achieve clustering partition(region partition)of radio map and coarse positioning of the target.According to the characteristics of radio map,a new clustering objective function is constructed to reduce the frequent singularity of k-means in the region partition process.In addition,support vector machine(SVM)is introduced to achieve coarse positioning to improve the positioning accuracy of k-means in the coarse positioning process.The experiment results manifest that the region partition and coarse positioning method based on the improved k-means and SVM can improve the fingerprint matching efficiency.Compared with the traditional k-means method,the method can significantly reduce the number of singular points and improve the coarse positioning accuracy,which improves the positioning accuracy.Finally,Kalman Filter which has better real-time performance is selected to achieve the positioning and tracking of motion target.For improving the accurancy of positioning and tracking,Weight K Nearest Neighbors(WKNN),the location fingerprint matching algorithm,is improved based on the change rule of received signal strength and transfer distance.The experiment results manifest that the improved WKNN combined with Kalman Filter can obtain better accurancy and stability in tracking motion target process.