文献类型：期刊文献

中文题名：模型匹配驱动的收集型传感网节能滤波机制

英文题名：Model-matching driven energy-saving filtering mechanism in gathering-oriented WSN

作者：黄如[1];张在琛[2];朱杰[3]

机构：[1]华东理工大学信息科学与工程学院,上海200237;[2]东南大学移动通信国家重点实验室,江苏南京210096;[3]上海交通大学电子工程系,上海200240

年份：2010

卷号：15

期号：5

起止页码：90

中文期刊名：电路与系统学报

外文期刊名：Journal of Circuits and Systems

收录：北大核心:【北大核心2008】；CSCD:【CSCD2011_2012】；

基金：国家自然科学基金项目资助(60802005);优秀青年教师科研基金项目资助(YH0157127)

语种：中文

中文关键词：传感器网络;数据收集;模型匹配;滤波机制;节能

外文关键词：WSN; data-gathering; model-matching; filtering mechanism; energy saving

摘要：论文面向传感器网络周期性收集中信息残缺流数据的统计特征,研究信源异构数据模型匹配驱动的节能滤波机制(MMF)。该机制在数据收集的作用阶段和信息容错方面区别于传统的面向传输过程的同构数据无损融合技术,能够在网络面向应用的前提下,针对异构信源采用有损融合方式来进一步降低网络整体能耗和传输时延。MMF的运作过程分别由模型建立的基本数据收集阶段和模型匹配驱动的自适应数据滤波阶段构成。首先依据半监督学习算法估计描述信源分布的高斯混合模型(GMM)参数,进而基于模型匹配程度来自适应控制网内数据通信频率,执行服务质量要求(QoS)约束下的数据收集有损融合。仿真实验表明,相比于一些经典的数据收集节能算法,本文提出的滤波机制能够在满足系统服务质量要求的前提下,通过提取信源异构流数据的统计冗余特征和驱动相应的模型匹配操作有效地抑制网内冗余数据传输次数和降低传输延时,最终实现健壮节能的传感器网络数据收集效果。
The paper addresses a model-matching filtering mechanism （MMF） for driving energy-saving periodic data-gathering in wireless sensor networks （WSN） via mining the statistical-characteristic of flow with imcomplete information at source. Distinguish from traditional lossless-fusion technologies adopted in the process of homogeneous-data transmission on the aspects of action-phases and fault-tolerability in gathering mechanism, our novel schemes focus on designing heterogeneous-data gathering mechanism at traffic source to further reduce total energy cost and transmission delay by using loss-fusion technology and the application-oriented trait of WSN. The whole operation processes of MMF are composed by two main stages. In the basic data-gathering stage, semi-supervised ｜earning method is adopted to estimate parameters of mixed GMM, which is applied to describe the statistical distribution characteristics of data block. Furthermore, model-matching driven filtering operation on traffic could be executed by adaptively controlling the frequency of communication operations and filtering the redundant loads in adaptive filtering stage. Simulation results show that MMF can achieve energy-saving and robust data-gathering on the premise of QoS requirement by extracting redundancy-attributes on heterogeneous flow data and driving corresponding model-matching filtering operation.