Journal of Beijing University of Posts and Telecommunications

  • EI核心期刊

Journal of Beijing University of Posts and Telecommunications ›› 2020, Vol. 43 ›› Issue (4): 95-100.doi: 10.13190/j.jbupt.2019-223

• REPORTS • Previous Articles     Next Articles

A Fingerprint Localization Method Based on Shapelet Algorithm

CHANG Zi-ying1, WANG Wen-han1, LI Tao1, LIU Fen1, CHEN Peng-peng1,2   

  1. 1. School of Computer Science and Technology, China University of Mining and Technology, Xuzhou 221116, China;
    2. Mine Digitization Engineering Research Center of the Ministry of Education, Xuzhou 221116, China
  • Received:2019-10-10 Published:2020-08-15

Abstract: Due to large influence of time and space on channel state information (CSI),the existing CSI-based indoor positioning technology is poor in robustness.Aiming at this problem,a fingerprint positioning method based on Shapelet algorithm is proposed.In the training phase,CSI is taken as the original location data,and the original data is processed and corrected by the 3-σ anomaly value processing method and the Kalman filter;then the fingerprint of each location is extracted and the fingerprint database is established by using the Shapelet algorithm;Finally,the fingerprint database is used to construct the Shapelet decision tree,and the decision tree classification is used to achieve more accurate positioning.Compared with the principal components analysis algorithm and,k-nearest neighbor algorithm,It is shown that the method has higher positioning accuracy and stable performance at different times,and the training set is smaller.

Key words: fingerprint location, channel state information, Shapelet algorithm, decision tree classification

CLC Number: