摘要
针对KNN指纹定位算法定位耗时长和基于K-Means聚类的KNN指纹定位算法定位精度不稳定的问题,本文提出了一种以接入点为离散点生成泰森多边形,利用泰森多边形对指纹聚类,然后使用最强接入点法确定移动节点的定位区域,最后通过动态KNN算法进行定位的指纹聚类定位算法。实验表明,该算法能有效缩短定位时间并提高定位精度,在接入点数量变化时表现出较好的定位性能,且在不同定位区域中性能具有较好的普适性。
To solve the problems of long position time-consuming in KNN fingerprint localization algorithm and unstable accuracy in K-Means clustering based localization algorithm,a novel fingerprint clustering localization algorithm is proposed. This algorithm considers APs as Voronoi diagram＇s generators to create Voronoi cells,uses these cells to cluster fingerprints of database and a method based on the biggest received signal strength to find out the positioning subarea of mobile nodes,and estimates the location of mobile node by automatic KNN algorithm. Experiment results reveal that this algorithm sharply reduces position time-consuming,improves the accuracy,and has a good performance when the quantity of AP changes. With the location area altering the performance still exists.
作者
吕娜
单志龙
张凡
余刘勇
LU Na , SHAN Zhilong , ZHANG Fan, YU Liuyong(Schoolof Computer Science, South China Normal University, Guangzhou 510631, Chin)
出处
《传感技术学报》
CAS
CSCD
北大核心
2017年第12期1941-1947,共7页
Chinese Journal of Sensors and Actuators
基金
国家自然科学基金项目(61671213
61370003)
广东省自然科学基金项目(2015A030313395)
广东省科技计划项目(2013B040401014)
