一种改进的Wi-Fi位置指纹室内定位算法  

An Improved Indoor Location Algorithm for Wi-Fi Position Fingerprint

作　　者：蒋盼盼[1] 林琼[1] 谢林蓉 JIANG Panpan;LIN Qiong;XIE Linrong(University of South China,Hengyang Hunan 421001,China;China University of Geosciences,Wuhan Hubei 430074,China)

机构地区：[1]南华大学,湖南衡阳421001 [2]中国地质大学,湖北武汉430074

出　　处：《通信技术》2021年第9期2132-2137,共6页Communications Technology

基　　金：湖南省教育厅科学研究项目(18C0421)。

摘　　要：经典的Wi-Fi位置指纹室内定位算法的在线匹配阶段通常采用加权K近邻算法(Weighted K-Near Neighborhood,WKNN),该算法定位移动对象时容易出现目标漂移,定位精度不高的情况。对此,本文提出了一种基于目标跟踪的加权K近邻算法和卡尔曼滤波的融合定位算法(Weighted K-nearest Neighbor Algorithm and Kalman Filter Fusion Localization Algorithm,WKNN-KF)。该算法充分考虑待定位点移动的连续性,首先利用加权K近邻算法对目标进行定位得到观测值,其次将观测值和卡尔曼滤波预估值进行加权求和,最后得到最优的估计坐标值。仿真实验结果发现,相比于加权K近邻算法,WKNN-KF定位算法对移动对象的运动轨迹定位更准确,算法的定位精度提高了45.7%,具有很好的推广应用价值。Classic indoor location algorithm of Wi-Fi location fingerprint uses the WKNN(weighted k-nearest neighbor)algorithm in the online matching stage,which makes the objects shifting and has a low precision in locating the moving object.In this paper,a WKNN-KF(Weighted K-nearest Neighbor Algorithm and Kalman Filter Fusion Localization Algorithm)based on target tracking is proposed.The algorithm fully considers the continuity of the movement of the point to be located,and first uses the weighted K-nearest neighbor algorithm to locate the target to obtain the observation value.Then,it performs a weighted summation of the observed value and the Kalman filter estimated value.Finally,the best estimated coordinate value is obtained.The simulation results indicate that compared with the weighted k-nearest neighbor algorithm,the WKNN-KF localization algorithm is more accurate for the moving object trajectory localization,and the positioning accuracy of the algorithm is improved by 45.7%,which has a fairly good practical value.

关 键 词：Wi-Fi位置指纹 加权K近邻算法 卡尔曼滤波 WKNN-KF 

分 类 号：TN92[电子电信—通信与信息系统]