An indoor positioning algorithm using joint information entropy based on WLAN fingerprint

Indoor positioning system in wireless local area network (WLAN) has become more and more popular accompanied by the popularization of GPS. While increasing numbers of access point (AP) enhancing the positioning accuracy little, the complexity increases a lot. To keep balance between the positioning accuracy and the complexity, we summarize three clustering methods named K-means, affinity propagation and fussy c means (FCM). In addition, a novel AP selection method named combination of information gain and mutual information entropy is proposed to decrease the computation cost. To get a higher positioning accuracy, we contrast two fine location methods named K nearest neighbor (KNN) and weighed K nearest neighbor (WKNN). The experiment results indicate that, the positioning accuracy within 2m is improved by using mutual information entropy and WKNN methods. By using cluster method and AP selection, we decrease the computation cost in the online phase a lot.