Fingerprint Indoor Localization Based On Data Mining

In recent years,with the rapid development of mobile Internet and the popularity of intelligent mobile terminals,location based services have received more and more attention.However,due to the limitations of indoor obstacles and narrow space,the traditional outdoor localization algorithms cannot be used in indoor localization.At present,the fingerprint indoor localization technology based on WIFI system has become a research hotspot because of its low deployment cost,flexible networking,easy to implement,convenient to expand and so on.Fingerprint indoor localization algorithm based on WIFI usually choose the RSSI(Received Signal Strength Indicator)received at each location area from APs(Access Points)as the localization characteristic fingerprint,using the mapping relation between RSSI and geographical position to realize localization.However,RSSI is easily affected by multipath,attenuation effect and environmental changes,which makes it difficult to construct a credible model for measuring the signal strength data,and this is a major challenge to improve the indoor positioning accuracy,which needs to an improvement with the new technology.Therefore,this paper introduces the theory of data mining into the indoor localization by consulting the relevant documents and analyzing data.The main work of this paper are as follows:(1)In order to analyze the characteristics of RSSI as a location fingerprint,we collect the RSSI samples in a real WIFI environment.And the method of combining theory with experiment is used to verify the RSSI of each AP in the environment.It is found that the RSSI of the signals sent by different APs in the WIFI environment have the uncertainty and repeatability.In order to accurately describe the relationship between RSSI and geographical position,it is necessary to collect as much as possible the different APs RSSI.But this would lead to an increase in the amount of computation of the localization algorithm.In addition,in the indoor environment,in order to ensure the coverage of WIFI network and data transmission quality,we tend to deploy a lot of APs.However,the RSSI from these APs have a high repeatability.Therefore,this paper proposes a fingerprint dimension reduction algorithm based on principle component analysis,to map the original high dimension fingerprint data to low dimension.By this algorithm,we reduced the data dimension and removed the redundant information between different APs at the same time.It is verified that the algorithm reduces the complexity of the algorithm and improves the localization efficiency.(2)In order to improve the accuracy of localization algorithm,this paper proposes a support vector regression localization model based on k layer parameters optimization.According to the nonlinear characteristics of RSSI,the current solution is to use the support vector machine based on kernel function to build positioning model,but the localization results of support vector machine is greatly affected by parameters.The traditional parameter optimization algorithm is inefficient and time-consuming.So by analyzing the reasons of poor efficiency of traditional algorithm,we use the idea of layering,choosing different search steps in different parameter intervals,it is verified the computational efficiency has been significantly improved.(3)Finally,this paper combines the fingerprint dimension reduction algorithm with the optimal parameters based support vector regression localization model,and simulated by data collected at real indoor environment.Experimental results show that compared with other data mining based localization algorithms like SVM,KNN and BPNN,the proposed algorithm shows superior performance in terms of localization accuracy.