WiFi Indoor Localization Algorithms Based On Multiple Selections Of Access Points

In recent years,location-based services(LBS)have been widely used in many aspects of people's lives.Indoor localization is an important support of LBS,so the research of indoor localization algorithm has aroused the interest of many researchers.However,due to the complexity and variability of indoor environment,the localization accuracy of indoor localization algorithm faces serious challenges.The main aim of this thesis is to improve the accuracy and stability of the indoor localization algorithm.In this paper,a Wi Fi indoor localization algorithm based on multiple selections of access points is firstly proposed.Firstly,this algorithm can effectively select a set of access points through multiple selection of Wi Fi access points,which has stable performance and strong resolution,and use this set of access points to represent the fingerprint characteristics of each location.Secondly,clustering the localization area based on the fingerprint characteristics of each location.Then,re-selecting the access points for each location cluster,and this set of access points can better show its characteristics for each location cluster.Finally,using C4.5 algorithm build a decision model for each location cluster.After modeling is completed,these models can be used for real-time localization.By multiple selection of access points and re-selection of access points,the location accuracy of the algorithm can be effectively improved.This paper further proposes a high performance fingerprint localization algorithm based on random forest.Aiming at the problem of weak generalization ability and overfitting of C4.5 algorithm,on the basis of Wi Fi indoor localization algorithm based on multiple selections of access points,this algorithm uses random forest algorithm instead of C4.5 algorithm to build decision model for each location cluster.The decision model not only has better localization accuracy,but also can avoid overfitting problem.In this paper,the corridor of area I,fourth floor,main building of Xidian university is taken as the experimental scene,and collecting Wi Fi data to verify the performance of the two localization algorithms.Two conclusions can be drawn: 1.The Wi Fi indoor localization algorithm based on multiple selections of access points has a good localization accuracy,which localization accuracy is more than 90% within the 2m localization error range,and the average localization error is 1.4959m;2.The high performance fingerprint localization algorithm based on random forest not only has good localization accuracy,but also has good algorithm stability,and its average localization error can be less than 1.3m.