AP Optimization Algorithm For Wi-Fi Indoor Localization Based On Multi-dimensional Feature Fuzzy

The indoor localization technology has important applications in many fields,however,with the wide deployment of wireless network Access Points(AP),the existing Wi-Fi fingerprint methods often neglect the existence of a large number of redundant AP and the difference of the position resolution of Received Signal Strength(RSS)features from different AP,which results in the low localization accuracy and efficiency.To address these problems,this thesis proposes an AP optimization algorithm for Wi-Fi indoor localization based on multi-dimensional feature fuzzy.The main contents are as follows.Firstly,a redundant AP reduction method based on fuzzy clustering is proposed.First,the multi-dimensional RSS features are extracted from the offline signal sample data,and the offline RSS feature matrix is constructed,which is preprocessed to obtain the standardized RSS feature data.Second,the one-dimensional and two-dimensional histogram estimation methods are used to calculate the Maximum Information Coefficient(MIC)between AP in the test environment ergodically,based on which the fuzzy equivalent matrix is constructed to characterize the correlation between AP.Finally,the clustering results of related AP can be obtained by fuzzy clustering under the action of cut matrix,which lays a foundation for the subsequent selection of non-redundant AP.Secondly,an AP position resolution calculation method based on fuzzy decision is proposed.First,the information gain ratio of each AP is calculated based on the offline RSS multi-dimensional features,and based on it the offline AP fuzzy membership is calculated.Second,the fuzzy relation equations are constructed by combining the offline AP fuzzy membership set with the fuzzy relational matrix of the offline RSS feature to solve out the fuzzy weight of each RSS feature.Finally,the multi-dimensional RSS features collected in the online phase are extracted to establish the fuzzy decision matrix about RSS features through fuzzy transformation,based on which the fuzzy relation equation are constructed and solved out again to obtain the online fuzzy membership of each AP,and which is defined as the position resolution of AP in the testing environment.Finally,the target position estimation under AP optimization is realized by combining Bayesian algorithm.In the online phase,on the basis of the completion of relevant AP clustering division,the AP with the highest online fuzzy membership is successively selected from each clustering,that is,AP with non-redundancy and higher position resolution are selected,and the target position estimation is completed by combining with the Bayesian positioning algorithm.The experiment results based on measured data show that the indoor positioning algorithm proposed in this thesis can not only improve the positioning accuracy but also reduce the positioning overhead in the online phase.