Research On Dynamic Fingerprint Window Model And Its Application Method Of Bluetooth Indoor Positioning

In recent years,the improvement of technologies such as the Internet of Things and information technology has increased people's expectations and demands of LBS(location based services)and LBS has become a hot spot in the industry.Among many indoor positioning methods,Bluetooth technology has become a research hotspot for indoor positioning with the advantages of low power consumption,easy deployment,and low cost.The fingerprint localization method has good stability,however,with the increase of positioning scene,the positioning system surfer more and more computational pressure.Also,the spatial difference of similar fingerprints will have a great impact on the accuracy.Focusing on the goal of reducing the operation cost of fingerprint positioning process and improving the positioning accuracy,this paper studies the fingerprint window which adapts to pedestrian motion characteristics and indoor space features,and proposes the application method of DFW(dynamic fingerprint window)model and application in Bluetooth fingerprint localization.The main research contents of this paper are as follows:(1)Propose a DFW model based on pedestrian motion analysis and indoor space characteristics.Establish the relationship between DFW and pedestrian movement speed and direction through indoor pedestrian movement analysis.Propose a DFW construction method based on the analysis of the influence of indoor space features on pedestrian movement.(2)Explain the construction method and usage method of DFW in Bluetooth indoor positioning.In order to increasing the speed of building a DFW in real-time positioning process,this paper proposes a fingerprint fast retrieval method based on the KD tree data structure.Considering that the KNN algorithm with fixed K value cannot get the optimal result,this paper proposed a dynamic K value determine method using similarity normalization index of fingerprints in DFW.(3)Carry out model experiments and evaluation.Experiments show that the DFW of this paper can effectively improve the accuracy of fingerprint location and the DF W-WKNN algorithm for DFW is more stable in positioning.