基于K-means聚类的蓝牙室内定位算法研究  

Research on Bluetooth Indoor Location Based on K-means Clustering

作　　者：余晔[1] Zhao Pengfei(Shanghai Surveying and Mapping Institute,Shanghai 200063,China)

机构地区：[1]重庆市勘测院

出　　处：《城市勘测》2019年第5期34-38,共5页Urban Geotechnical Investigation & Surveying

摘　　要：提出一种基于K-means聚类的蓝牙指纹室内定位算法,首先利用K-means聚类对定位区域的指纹数据库进行划分得到不同子区域的类中心以及每个类的指纹参考点;然后,在线定位时将接收到的蓝牙信号强度与不同的类中心进行比较,匹配到对应的子区域中。其次,本文提出了一种基于位置加权k-最邻近(Weight K-Nearest Neighbor,WKNN)定位算法,其中WKNN中的权重选取是利用多次定位结果的方差决定。实验结果表明利用区域划分和基于位置方差加权的WKNN算法可以有效提高定位精度。A Bluetooth fingerprint indoor location algorithm based on K-means clustering is proposed in this paper. Firstly,the fingerprint database of the location area is partitioned by K-means clustering to get the class centers of different sub-regions and the fingerprint of reference points of each class. Then,in on-line phase the received Bluetooth signal strength is compared with the different class centers. Secondly,a WKNN localization algorithm based on location weighting is proposed,in which the weight selection in WKNN is determined by the variance of multiple positioning results. Finally,the experimental results show that the location accuracy can be effectively improved by using region partition and WKNN algorithm based on location variance weighting.

关 键 词：室内定位 蓝牙指纹 聚类 位置加权 

分 类 号：TN911.2[电子电信—通信与信息系统]