计算机工程

• 移动互联与通信技术 • 上一篇    下一篇

基于StOMP算法的WSN压缩感知数据重构

黄志清 1,2,张严心 3,李梦佳 1,2,成志鹏 3   

  1. (1.北京工业大学 软件学院,北京 100124; 2.北京市物联网软件与系统工程技术研究中心,北京 100124; 3.北京交通大学 电子信息工程学院 先进控制系统研究所,北京 100044)
  • 收稿日期:2016-07-15 出版日期:2017-09-15 发布日期:2017-09-15
  • 作者简介:黄志清(1970—),男,副教授、博士,主研方向为无线传感器网络、物联网、软件定义网络;张严心(通信作者),博士;李梦佳、成志鹏,硕士。
  • 基金项目:
    国家发改委项目(Q5025001201502);中央高校基本科研业务费专项资金(W16JB00340)。

WSN Compressed Sensing Data Reconstruction Based on StOMP Algorithm

HUANG Zhiqing 1,2,ZHANG Yanxin 3,LI Mengjia 1,2,CHENG Zhipeng 3   

  1. (1.School of Software,Beijing University of Technology,Beijing 100124,China; 2.Beijing Engineering Research Center for IoT Software and Systems,Beijing 100124,China; 3.Advanced Control Systems Laboratory,School of Electronic and Information Engineering,Beijing Jiaotong University,Beijing 100044,China)
  • Received:2016-07-15 Online:2017-09-15 Published:2017-09-15

摘要: 分段正交匹配追踪算法(StOMP)运算速度快、计算量小,适用于无线传感器网络(WSN)压缩感知数据重构。为此,分析并研究StOMP算法的门限阈值选取对WSN压缩感知数据重构精度的影响,提出一种StOMP算法门限阈值的自适应调整方法。基于比例-积分-微分方法的思想,根据StOMP算法的当次重构误差计算门限阈值的调整值,并使用调整后的门限阈值重新进行数据重构,重复该过程以提高重构精度。实验结果表明,该方法能快速找到满足误差要求的门限阈值,与采用固定门限阈值的调整方法相比,重构精度更高。

关键词: 无线传感器网络, 压缩感知, 数据重构, 分段正交匹配追踪算法, 比例-积分-微分方法

Abstract: Stagewise Orthogonal Matching Pursuit(StOMP) algorithm has rapid computing speed and small amount of calculation quantity,and is thus highly suited to the reconstruction of Wireless Sensor Network(WSN) Compressed Sensing(CS) data.The influence of threshold selection of StOMP algorithm on the reconstruction accuracy of CS data in WSN is analyzed and studied.And this paper proposes an adaptive adjustment method of threshold value in the StOMP algorithm.Based on the idea of Proportional-Integral-Derivative(PID) method,the adjustment value of the threshold is calculated according to the reconstruction error.Reconstruct the data with the new threshold,and this process is repeated to improve the accuracy of reconstruction.Experimental result shows that this method can quickly find the threshold value which meets the error requirement and has distinctly improved reconstruction precision compared with the adjustment method of fixed threshold value.

Key words: Wireless Sensor Network(WSN), Compressed Sensing(CS), data reconstruction, Stagewise Orthogonal Matching Pursuit(StOMP) algorithm, Proportional-Integral-Derivative(PID) method

中图分类号: