基于观测节点不等概随机投影的压缩感知.pdfVIP

第 17 卷第 1 期 信息工程大学学报 Vo l. 17 No. 1 2016 年 2 月 Journal of Information Engineering University Feb.2016 DOI: 10. 3969/j. issn. 1671 -0673.2016.0 1. 003 基于观测节点不等概随机投影的压缩感知 李丹1 ，李鸥1 ，张翔2 (1.信息工程大学，河南郑州 450001 ; 2.78006 部队，四川成都 610041 ) 摘要:基于空间相关性的压缩感知是无线传感器网络的一个重要研究内容，采用随机选取节点 的方法构建观测矩阵可以有效降低观测节点的能量损耗。提出一种基于不等概随机投影的节 点选择策略，每个节点根据采集数据和门限的比较结果选择高概率或低概率，并根据能量均衡 策略对其进行调节后以该概率发送数据至融合中心。仿真结果表明，与等概率随机投影相比， 文章提出的策略在不增加额外能量损耗的前提下，降低了整个区域重构误差。 关键词:无线传感网;压缩感知;稀疏观测矩阵;不等概随机投影;能量均衡 中图分类号:TN92 文献标识码:A 文章编号: 1671 -0673 (2016 )01 -0011 -06 Compressed Sensing Based on Random Projection of Observation Nodes with Unequal Probabilities 2 LI Dan , LI Ou , ZHANG Xiang (1. Information Engineering University , Zhengzhou 450001 , China; 2. Unit 78006 , Chengdu 610041 , China) Abstract: Compressed sensing based on temporal and spatial correlation is one important research field in wireless sensor network ( WSN ). Making the measurement matrix through random nodes se- lection can effectively reduce the energy consumption of observation nodes. A strategy to select nodes based on random projection with unequal probabilities is proposed. Each node chooses a high- er probability or a lower one according to the comparison between its measurement and a threshold. It revises the probability according to an energy balance strategy and sends its observation to the fu- sion center with the revised probability. Simulation results demonstrate that compared with random projection with equal probability , the proposed