摘要
为解决传统加权K最近邻算法(WKNN,Weighting K-Nearest Neighbor)定位方法中选取K值存在局限性影响定位精度的问题,提出了一种改进型几何聚类指纹室内定位方法。该方法首先利用网格分布在定位区域构建指纹点几何位置分布,采集指纹点接收信号强度(RSS,Received Signal Strength)和位置信息,建立指纹定位数据库;然后,利用支持向量机分类算法在解决高维度和非线性问题上的优势选取定位点的多个近邻指纹点,根据对定位贡献度的大小筛选近邻指纹点并构建几何聚类定位区域;最后利用WKNN算法进行定位。实验结果表明,提出的方法解决了传统WKNN方法中多边形定位区域在K值选取存在局限性的问题,具有更高的定位精度和工程实用性。
In order to solve the problem of limitation about how to choose the value K of Weighting K-Nearest Neighbor algorithm(WKNN,Weighting K-Nearest Neighbor) in indoor positioning, an improved indoor positioning method based on geometric clustering fingerprinting is proposed. Firstly, the grid distribution is used to establish the geometric distribution of fingerprint points, and the data of the Received Signal Strength(RSS) and location can be collected, then the positioning database can be built. Secondly, the support vector machine algorithm is used to obtain the neighboring fingerprint points about positioning point due to the advantages in solving the problems of nonlinearity and high dimensions, and the geometric clustering positioning area can be constructed by the neighboring fingerprint points according to the contribution to positioning point. Finally, the results can be got by WKNN algorithm. The experimental results show that the proposed method solved the problem of limitation in selecting the value K of WKNN algorithm and polygonal positioning area, it has higher precision and better engineering practicality than traditional methods.
作者
李石荣
符茂胜
张峰辉
何富贵
LI Shirong;FU Maosheng;ZHANG Fenghui;HE Fugui(School of Electronics and Information Engineering,West Anhui University,Lu’an 237012,China)
出处
《四川理工学院学报（自然科学版）》
CAS
2019年第4期28-33,共6页
Journal of Sichuan University of Science & Engineering(Natural Science Edition)
基金
国家自然青年科学基金项目(61702375)
安徽省自然科学基金项目(1908085MF213)
安徽高校自然科学研究重大项目(KJ2015ZD44)
皖西学院青年项目(WXZR201806)
安徽高校自然科学研究项目(KJ2019A0631)
关键词
K最近邻算法
室内定位
几何聚类
接收信号强度
支持向量机
贡献度
K-Nearest Neighbor algorithm
indoor positioning
geometric clustering
Received Signal Strength(RSS)
Support Vector Machine (SVM)
contribution
