Analysis of K-Means algorithm on fingerprint based indoor localization system

Abstract

The collected fingerprints at the anchors in indoor localization system are clustered with corrected K-Means algorithm in order to reduce the computational complexity in the online localization phase. When the WLAN indoor environment contains enough access points (APs), every anchor's fingerprint may have too many different dimensions. Therefore these fingerprints should be principal component analysis (PCA) and set dimension's property dynamically when clustering. The up number limit of clusters for common fingerprint database is provided. And the optimized cluster number within the up number limit and default dimension setting are provided simultaneously.