Research On Indoor Positioning Based On WiFi Fingerprint Location Algorithm

In recent years,with the development of indoor positioning technology,indoor positioning by providing location information data for the retail,manufacturing,robotics and other industries,has become an important foundation for the era of the Internet of Things.The indoor positioning technology combines people and objects with data information,so that it can be searched and positioned like the image information,realizing the Internet of Everything.There are many existing indoor positioning technologies,and different positioning technologies are different in positioning accuracy,positioning cost,and application range.With the popularity of WLAN,WLAN-based indoor positioning technology has become a hot topic in recent years due to its wide range of applications,low positioning cost,and high positioning accuracy.This dissertation conducts in-depth investigation and research on WiFi-based fingerprint location technology,and solves how to divide the location area step by step in a large-scale indoor scene,and narrow the target to a specified area,thereby improving positioning accuracy and reducing algorithm complexity.In this dissertation,neural network,self-encoder,K-means,KFCM and other methods are applied,and the feasibility is verified under the designed positioning system.The main research work of this dissertation is as follows:First,the indoor floor location is studied,and a floor classification algorithm based on neural network is proposed.WiFi has a wide coverage,and its positioning accuracy determines that it is not suitable for indoor positioning in a small range.In a complex,floor positioning solves the coarse positioning of the target,thereby reducing the search range of the fingerprint library.In this dissertation,the positioning feature is extracted by the self-encoder,and the floor is verified by the Softmax classifier,which verifies that the fingerprint processed by the self-encoder has better performance in the floor positioning.Secondly,the use of clustering algorithm in fingerprint location is studied,and a clustering method based on reference point location is proposed.The WiFi strength information is susceptible to interference,large fluctuations,and prone to outliers,resulting in the traditional clustering method based on fingerprint intensity vector is not stable enough.Considering that the reference point information also contains position coordinates,and the position coordinates are stable and uniform,therefore,the K-means clustering method based on position coordinates is adopted in this dissertation.The introduction of the kernel method can divide the linear inseparable fingerprint data and has higher inclusiveness to the outliers.Therefore,this dissertation also introduces the KFCM clustering technique.The experimental results show that the clustering of fingerprint database can effectively improve the positioning accuracy.Thirdly,the AP selection algorithm is improved,and a server-based WiFi positioning system is designed and implemented.In a large range of indoor scenes,the number of APs is large,and the coverage range is different.It is not feasible to perform AP selection when building a fingerprint database.Therefore,this dissertation proposes to move the AP selection to the real-time positioning intensity vector.Finally,the above method is tested in the positioning system designed and implemented in this dissertation,and the effectiveness of the method is verified.