基于KNN算法的可变权值室内指纹定位算法  

Variable-weight Indoor Fingerprinting Localization Algorithm Based on KNN Algorithm

作　　者：戴志诚[1] 李小年 陈增照[1] 何秀玲[1] DAI Zhicheng;LI Xiaonian;CHEN Zengzhao;HE Xiuling(National Engineering Research Center for E-Learning,Central China Normal University,Wuhan 430000,China)

机构地区：[1]华中师范大学国家数字化学习工程技术研究中心

出　　处：《计算机工程》2019年第6期310-314,共5页Computer Engineering

基　　金：国家重点研发计划(SQ2018YFB10006,2017YFB1401303)

摘　　要：针对基于静态权值的室内指纹定位算法存在定位精度低、定位结果不稳定以及环境适应性差等问题,提出一种以欧氏距离为权值参考的可变权值室内指纹定位算法。该算法分为离线采样阶段和在线定位阶段。离线采样阶段对接收信号强度指示(RSSI)值进行高斯滤波去噪构建指纹库。在线定位阶段引入权值指数α、β,分别以RSSI、欧氏距离为权值参考计算最近邻点及其加权质心,得出待测节点的坐标。实验结果表明,相比KNN和RW算法,该算法定位精度高,其平均误差为0.965m,且定位误差波动小。To address the problems of low positioning accuracy,unstable positioning results and poor environmental adaptability of indoor fingerprinting localization algorithm based on static-weight,this paper proposes a variable-weight indoor fingerprinting localization algorithm based on Euclidean distance.The algorithm is divided into offline sampling phase and online localization phase.In the offline sampling phase,the algorithm performs Gaussian filter denoising on the Received Signal Strength Indication (RSSI) value to construct a fingerprint library.The weight of indexs,α and β,are introduced in the online localization phase,and the RSSI value and Euclidean distance are referenced as weight to figure out the nearest neighbor points and their weighted centroid.Experimental results show that compared with KNN algorithm and RW algorithm,the proposed algorithm has a higher positioning accuracy with the mean error of 0.965 m,and errors are less volatile.

关 键 词：接收信号强度指示 KNN算法 可变权值 加权质心 指纹定位 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]