MR Address Fingerprint Location Technology Based On Position Sensing Data Fusion

With the development of modern wireless communication network technology,location-based services(LBS)have been deeply integrated into people's lives.According to statistics,80% of people's activities are carried out indoors,such as market,hospital,home,fire prevention and rescue and other industry needs,so it is of great significance to study indoor positioning projects.The location technology based on MR address fingerprint and pedestrian dead reckoning(PDR)has become a research hotspot in the field of location technology because of its fast,accurate,high accuracy and low cost.However,the disadvantage of MR fingerprint positioning technology is that it presents jumping positioning,and the positioning accuracy is easily affected by interference signals.The other PDR positioning technology has the problem of error accumulation,which leads to the decline of positioning accuracy.Therefore,this paper studies a positioning method which combines MR fingerprint positioning and PDR positioning technology,and combines with adaptive filter for position prediction.In order to further improve the positioning accuracy,this paper provides observation correction values through MR fingerprint positioning and correction parameters for PDR state prediction.Firstly,according to the combination characteristics of wireless signal strength in the spatial grid,this paper studies the establishment of MR address fingerprint database,and makes theoretical analysis and simulation comparison of nearest neighbor method(NN),k-nearest neighbor method(KNN),weighted k-nearest neighbor method(WKNN)and BP neural network algorithm in the positioning stage,and selects WKNN algorithm as the application algorithm of MR fingerprint positioning.Secondly,the paper studies the pedestrian dead reckoning method based on sensor data,including the acceleration sensor data and direction sensor data acquisition,signal denoising processing,pedestrian progress,step size and direction,positioning coordinates and so on.Finally,in order to solve the problem of large cumulative error of PDR localization algorithm and the drift of MR fingerprint localization,this paper uses extended Kalman filter(EKF)algorithm and particle filter(PF)algorithm to fuse MR localization and PDR localization algorithm for localization prediction.In this paper,MR fingerprint positioning,PDR positioning,extended Kalman filter fusion positioning method and particle filter fusion positioning method are simulated.Through the comparative test,it is concluded that the positioning accuracy of extended Kalman filter and particle filter fusion method is greatly improved compared with the traditional MR positioning algorithm and PDR positioning algorithm.