基于稀疏表示的室内指纹定位算法  

An indoor fingerprint location algorithm based on sparse representation

作　　者：陈仲民[1] 樊恒[2] 向金海[1] 邓君丽[1] 

机构地区：[1]华中农业大学信息学院,湖北武汉430070 [2]华中农业大学工学院,湖北武汉430070

出　　处：《武汉大学学报（工学版）》2016年第3期465-468,475,共5页Engineering Journal of Wuhan University

基　　金：中央高校基本科研业务费专项资金资助项目(编号:2662014PY052);博士科研启动专项(编号:2662014BQ083)

摘　　要：为解决室内实时定位中精度不高的问题,提出了一种基于稀疏表示的室内指纹定位算法.针对传统的指纹数据库匹配算法的不足,将待测点的位置估计看作多分类问题.首先在室内区域选择若干个参考点,多次测量参考点的WiFi信号强度,构建稀疏数据字典.通过稀疏表示的方法,用参考点的指纹矢量对待测点处的指纹矢量进行重构,计算重构误差并根据其对待测点位置进行估计.实验结果表明,与传统SVM定位方法相比,该算法的定位精度有明显提高.To solve the problem of low accuracy in real-time location, an algorithm based on sparse representation is proposed. For the disadvantages of traditional location methods, we view the location as multipie classification. Firstly we select some reference nodes in indoor area, and measure the WiFi signal strength of reference nodes for multiple times to construct the sparse dictionary. Then the fingerprint of test nodes can be reconstructed with sparse dictionary. According to the reconstruction error, we can classify the test nodes and estimate its location. Experimental results demonstrate that our algorithm outperforms SVM location methods.

关 键 词：稀疏表示 WIFI 室内定位 指纹算法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]