Research On Fingerprint AP Selection Strategy Of Indoor WiFi Location

In the past 100 years,various systems such as communication,broadcast:ing and positioning and navigation,which have been established by using radio technology,have become the most important technological infrastructure in human society,ubiquitous,omnipotent and timeless.In order to make up for the lack of positioning accuracy of the satellite positioning and navigation system in the indoor environment and the difficulty in guaranteeing the quality of service,it is a major supplement or alternative to use the widely deployed WiFi wireless network to assist in providing indoor positioning services.The indoor positioning method based on WiFi location fingerprint has the characteristics that the signal fingerprint feature is not easily afifected by the indoor signal propagation path,and has attracted the attention of researchers.Access points(APs)densely deployed in indoor environments increase the dimension of WiFi location fingerprint features,improve location resolution,and improve positioning accuracy to a certain extent.However,the consequent increase in the size of the WiFI location fingerprint library has created significant challenges in computing,storage,and management.In the indoor environment,the characteristics of the WiFi location fingerprint are reduced by selecting the AP access point related strategy,and the problem of reducing the positioning accuracy by the effective auxiliary method is studied.The research on the WiFi location fingerprint indoor positioning system has certain Theoretical guidance and application value.Based on statistical theory and machine learning methods,this paper adopts the research method of system modeling.The main research work is as follows:(1)For the AP location point in the indoor environment,the WiFi location fingerprint feature dimension is too large.First,the capacity of the reference point sample is minimized by optimizing the sampling function(OC function).The second is to propose an AP selection strategy based on chi-square test,and carry out comparative analysis and research.(2)Aiming at the problem of lowering the positioning accuracy caused by WiFi location feature fingerprint dimensionality reduction,an auxiliary fingerprint direction compensation method based on auxiliary fingerprint database is proposed.By establishing an auxiliary fingerprint database offline,obtaining direction compensation labels and compensation errors,using the Random Forest(RF)method to learn the classification model and positioning the positioning points.(3)According to the typical experimental building corridor environment,by collecting a large amount of data,the four AP selection strategies are combined with the positioning calculation of auxiliary fingerprint direction compensation.The results show that in the AP selection problem,the four AP selection methods increase with the number of APs,and the average positioning error shows a downward trend.The AP selection strategy based on the chi-square test has a faster decline rate.When the number of APs reaches 5 The positioning error tends to be flat,indicating that the AP selection based on the chi-square test is feasible.However,the average positioning error(0.549m)of all AP methods has decreased.Therefore,combined with the auxiliary fingerprint direction compensation strategy,the loss ratio has dropped from 39.25%to 13.37%.