作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程 ›› 2018, Vol. 44 ›› Issue (7): 109-113,120. doi: 10.19678/j.issn.1000-3428.0046940

• 移动互联与通信技术 • 上一篇    下一篇

基于支持向量机的混合相似度室内指纹定位算法

施涛涛 a,b,卢先领 a,b,于丹石 b   

  1. 江南大学 a.轻工过程先进控制教育部重点实验室; b.物联网工程学院,江苏 无锡 214122
  • 收稿日期:2017-04-24 出版日期:2018-07-15 发布日期:2018-07-15
  • 作者简介:施涛涛(1992—),男,硕士研究生,主研方向为无线传感网络、室内定位技术;卢先领,教授、博士;于丹石,工程师。
  • 基金资助:

    江苏省产学研联合创新资金前瞻性联合研究项目(BY2014023-31);江苏省“六大人才高峰”项目(WLW_007)。

Mixed Similarity Indoor Fingerprint Location Algorithm Based on Support Vector Machine

SHI Taotao a,b,LU Xianling a,b,YU Danshi b   

  1. a.Key Laboratory of Advanced Process Control for Light Industry,Ministry of Education; b.School of Internet of Things Engineering,Jiangnan University,Wuxi,Jiangsu 214122,China
  • Received:2017-04-24 Online:2018-07-15 Published:2018-07-15

摘要:

传统室内指纹定位算法因参考点匹配精度低,导致定位误差大。为此,提出基于支持向量机(SVM)的混合相似度加权K近邻算法SVM-MWKNN。在离线阶段,将已采集的接收信号强度值进行去干扰处理后,对指纹库作网格划分,然后使用SVM实现网格学习;在定位阶段,使用SVM进行 网格寻找,确定待定位点指纹所属网格,然后求出k个最大相似度值作为权值以实现定位。实验结果表明,与已有的位置指纹定位算法相比,该算法通过建立多相似度指标,可以有效提高数据利用率,减少定位误差,定位精度提高达45%。

关键词: 指纹定位, 支持向量机, 网格法, 局部相似度, 整体相似度

Abstract:

The location error of traditional indoor fingerprint location algorithm is high because of the low precision of reference point matching.Therefore,a mixed similarity weighted K nearest neighbor algorithm based on Support Vector Machine (SVM) is proposed.In the off-line phase,after the received signal strength values are processed,the fingerprint library is meshed,and then SVM is used to study the grid.In the positioning stage,the grid is searched by SVM to determine the grid of the pending point fingerprint,and then the maximum k similarity value is obtained as the weight value to realize the positioning.Experimental results show that,compared with the existing fingerprint location algorithm,the proposed algorithm can effectively improve the data utilization,reduce the location error and improve the positioning accuracy by 45%.

Key words: fingerprint location, Support Vector Machine(SVM), grid method, local similarity, overall similarity

中图分类号: