基于K均值聚类算法的位置指纹定位技术

Location fingerprint positioning technique based on K-means clustering algorithm

为了减小位置指纹定位算法的计算量,提出一种基于K均值聚类分析的位置指纹定位算法。通过对指纹数据库进行K聚类分析,形成聚类索引,定位时通过查询聚类索引来缩小指纹库查询空间。利用改进后的算法进行室内定位实验,并将其与K近邻法进行对比测试。实验结果表明,改进后的定位算法有效减小了定位过程的计算量,而且还能保证定位精度,在短距离范围内定位平均误差可限制在2m以内。

In order to reduce the computational complexity of location fingerprint positioning algorithm,a location fingerprint positioning algorithm based on K-means clustering is proposed. By making a K-means clustering analysis of the fingerprint database,the cluster index is formed for reducing the search space when positioning. Using the improved algorithm makes location experiment,and comparing with K nearest Neighborhood algorithm.The experimental result shows that the computational complexity is reduced and the better location performance can be achieved by improved algorithm,and this algorithm makes the average error to be limited within 2 m in short range.