A novel adaptive weighted K-nearest neighbor positioning method based on omnidirectional fingerprint database and twice affinity propagation clustering

Abstract

Human body has a great influence on Wi-Fi signal propagation. Therefore, we present a novel adaptive weighted K-nearest neighbor (KNN) positioning method based on omnidirectional fingerprint and twice affinity propagation clustering considering user’s orientation. Firstly, an improved fingerprint database model named omnidirectional fingerprint database (ODFD) is proposed, which includes the position, orientation and the sequence of mean received signal strength indicator at each reference point. Secondly, affinity propagation clustering (APC) algorithm is introduced for clustering on the offline stage based on the hybrid distance, which is the fusion of signal-domain distance and position-domain distance. Finally, adaptive weighted KNN algorithm based on APC is proposed. KNN method is exploited to obtain K initial reference points (RPs), then all of them are clustered by APC algorithm based on RPs’ position-domain distances. The most probable cluster is reserved by the comparison of RPs’ number and signal-domain distance between cluster center and test point. The weighted average coordinate value of residual RPs in the remaining cluster can be estimated. We have implemented the proposed method with the mean error of 2.2 meters, the root mean square error of 1.5 meters. Experimental results show that our proposed method outperforms traditional fingerprinting methods.