Research On Indoor Positioning Algorithm Based On Location Fingerprint

With the emergence of large-scale commercial center with integration of shopping, entertainment, dining, and leisure, the demand of indoor positioning and navigation is increasing, and this area has become a hot research topic. Indoor navigation first solves people's own real-time positioning problem. With the popularity of Wi Fi technology and evolving applications, indoor positioning technology based on WiFi gets the attention of scholars and engineering because of its unique advantages. But the existing WiFi positioning schemes will be affected by interfere in the indoor environment, their positioning precision and efficiency are not high, unable to meet the requirements of indoor localization project. So this thesis aims to design an indoor positioning solution that could satisfy the requirement of engineering.Firstly, the key factors affecting indoor positioning accuracy and efficiency are investigated. Fingerprint positioning method which is optimum in WiFi-based positioning is studied, and we have analyzed the process of the fingerprint positioning method. The matching algorithm and external factors that affect the fingerprint positioning accuracy and efficiency have been analyzed from both a qualitative approach and a quantitative approach.Secondly, this thesis introduces the idea of combining the clustering method with fingerprint positioning to improve the efficiency of the program location, using Kmeans clustering for data matching before pretreatment. Fingerprint database are generally large, traditional fingerprint location directly uses the fingerprint database to match each data record test points with RSSI(received signal strength indication). This matching process normally takes a long time, and it will have a negative impact of designing efficient fingerprint localization algorithm. Based on the study of establishing the fingerprint database and analyzing existing clustering, we design a method to introduce Kmeans clustering method to preprocess the fingerprint database. The effectiveness of our method to preprocess fingerprint database using clustering validity Kmeans pretreatment is demonstrated through Matlab simulation.Finally, this thesis has improved the defects of Bayesian probability algorithm. We designed a kind of improved localization algorithm based on Bayesian probability algorithm. And the positioning accuracy of our improved algorithm is verified by Matlab simulation. We first analyze the major reason that causes the matching algorithm to have the low accuracy. Based on this, we have introduced some improvement. Moreover this thesis adopts the strongest AP principle to improve the Bayesian probability algorithm, solve the problem that the Bayesian algorithm has a big positioning error in certain circumstances. Finally, in order to improve positioning accuracy, as well as verify how the modified Bayesian positioning algorithm has improved the positioning accuracy, this thesis has carried out simulation from different angles.