Multi-method Fusion Node Positioning Technology Based On Location Fingerprint

With the rapid development of the Internet and mobile intelligent terminals,location-based services are also favored by the public.The economic,efficient and accurate location services have broad prospects for development.At present,outdoor GPS positioning technology can meet people's demand for location,and indoor node positioning technology is inaccurate due to the signal is affected by environmental interference.Therefore,improving the positioning accuracy of indoor nodes has become a hot topic in current research.The research topic of this thesis is multi-method fusion node location technology based on location fingerprint.The research content is divided into two parts,namely,the location model based on region partitioning and adaptive dynamic weighting KNN,and the fusing technology of WiFi and magnetic signals.Specifically completed the following work:1.In wireless network environment,owing to the received signal intensity is easily affected by the environment and fluctuates,an inaccurate fingerprint database is constructed,which results in inaccurate location.This thesis based on filtering,fingerprint classification and improved nearest neighbor algorithm,a node location model is proposed,which combines region division and adaptive dynamic weighted KNN.Firstly,Kalman filter and variance filter are combined to optimize the sample data of received signal strength to form a position fingerprint database;secondly,the region to be located is divided according to the maximum signal strength value of the reference point,and the fingerprint classification is completed;and then,the weight coefficients of each reference point are calculated according to the received signal strength value,and the weight vectors in the region are obtained,which are adaptively selected with Euclidean distance.The KNN algorithm is improved by selecting K adjacent reference points,and the coordinate estimation of unknown nodes is finally realized.The experimental results show that the proposed method can effectively improve the position accuracy.2.To solve the problem of single node location feature,this thesis presents a location method combining WiFi and geomagnetism.Firstly,the non-linear mapping between geomagnetic signal intensity and physical location is made by using BP neural network to obtain the location estimation based on geomagnetic field.Then,the constrained Lagrange equation is established by fusing the WiFi estimation region.According to the results of WiFi and geomagnetic location,the initial coordinates of unknown nodes are calculated respectively,and the gradient approximation method is introduced to iteratively update the step size,and the initial coordinates are corrected.Finally,getting the final coordinates.The experimental results show that the average error of the proposed method is 15.8% and 10.9% higher than that of the single WiFi and geomagnetic positioning,under 2-m accuracy is 83%,which is 9.2% higher than that of the existing fusion methods.In the actual environment,the positioning method proposed in this thesis can not only improve the positioning accuracy,but also have good stability.