Journal of South China University of Technology (Natural Science Edition) ›› 2017, Vol. 45 ›› Issue (10): 87-92,99.doi: 10.3969/j.issn.1000-565X.2017.10.012

Previous Articles     Next Articles

Design of EKNN Algorithm for Regional WiFi Localization

FU Yu-li YANG Shuai CHEN Pei-lin HUANG Zhi-jian TANG Jie   

  1. School of Electronic and Information Engineering,South China University of Technology,Guangzhou 510640,Guangdong,China
  • Received:2016-11-23 Revised:2017-03-22 Online:2017-10-25 Published:2017-09-01
  • Contact: 傅予力(1959-)，男，教授，博士生导师，主要从事智能信息处理研究． E-mail:fuyuli@scut.edu.cn
  • About author:傅予力(1959-)，男，教授，博士生导师，主要从事智能信息处理研究．
  • Supported by:
    Supported by the National Natural Science Foundation of China(61471174)

PDF

316

Abstract: Since WiFi is widely used,it is selected as one of the popular technologies for indoor positioning.The positioning accuracy and speed have always been hot research issues.In this paper,an evidence K-nearest neigh- bor(EKNN) algorithm is proposed for the localization of the region of interest(ROI).In the algorithm,first,by taking a received signal strength indicator as the fingerprint,a wireless fingerprint database is established as a recognition class in each region.Then,based on the evidence theory,the neighbor evidences within each class are combined and the combined results among the classes are fused.Finally,the ROI class of the target is deter- mined,and a precise positioning is performed within this class.As compared with other algorithms,the EKNN al- gorithm can achieve a recognition rate of 97% at best and a maximum positioning error of about 2. 2 meters as well as a great positioning efficiency improvement.

Key words: WiFi positioning, region of interest, evidence K-nearest neighbor, wireless fingerprint database, received signal strength indication

CLC Number: