基于改进K-means算法的WSN簇头节点数据融合

发布时间：2018-01-14 21:34

本文关键词：基于改进K-means算法的WSN簇头节点数据融合　出处：《农业机械学报》2015年S1期 　论文类型：期刊论文

【摘要】：无线传感器网络数据融合能够减少节点能耗、延长网络生命周期,近年来受到了广泛关注。已有的应用于农业监测的空间数据融合算法多采用取平均值等方法将一定区域内监测到的数据融合成一个值。而农田环境监测具有监测范围广、监测点多、监测数据量大的特点,监测数据间除了冗余性还具有差异性,因此数据融合应该在消除冗余的同时保留数据的差异。针对农业监测的这一特点,提出在簇头节点应用聚类算法进行空间数据融合,通过聚类减少数据发送量,降低能耗;同时将差异较大的参量聚类到不同类别中以保留数据间的差异。此外,还提出了一种应用于WSN簇头节点的自适应改进K-means聚类算法,仿真结果表明,所提算法融合后的数据上传量比没有融合减少41.19%,消除了数据冗余;算法融合前后最大误差低于取平均值法误差的36%,保留了数据差异性。在没有明确误差要求时,该算法能够在尽量减少数据上传量的同时保持相对误差低于10%,避免了因聚类个数不当引起的巨大误差。而在有具体误差要求时,该算法融合前后的绝对误差严格低于要求误差。
[Abstract]:Wireless sensor network data fusion can reduce node energy consumption and prolong network life cycle. In recent years, the spatial data fusion algorithms used in agricultural monitoring have been widely concerned. Most of the spatial data fusion algorithms, such as average value, are used to fuse the monitored data into a single value. However, the monitoring of farmland environment has the characteristics of monitoring. A wide range. There are many monitoring points, large amount of monitoring data, monitoring data is also different in addition to redundancy, so data fusion should eliminate redundancy at the same time to retain the differences of data. In view of this feature of agricultural monitoring. A clustering algorithm is proposed for spatial data fusion in cluster head nodes, which can reduce the amount of data transmission and reduce energy consumption. At the same time, the parameter clustering with large difference is applied to different categories to preserve the differences between the data. In addition, an adaptive improved K-means clustering algorithm for WSN cluster heads is proposed. The simulation results show that the amount of data uploaded by the proposed algorithm is 41.19 less than that without fusion, and the data redundancy is eliminated. The maximum error of the algorithm before and after fusion is lower than the average error of 36 points, which preserves the difference of data. The algorithm can reduce the amount of data uploads as much as possible while keeping the relative error below 10, avoiding the huge error caused by the improper number of clustering. The absolute error of the algorithm before and after fusion is strictly lower than the required error.
【作者单位】：中国农业大学信息与电气工程学院;
【基金】：国家自然科学基金资助项目(31371531)
【分类号】：TP202;TP212.9;TN929.5
【正文快照】： 引言随着现代农业的不断发展,用无线传感器网络(Wireless sensor network,WSN)对农业产地环境进行监测已经成为开展精细农业、提高农作物产量的重要方法之一[1-4]。在农业产地环境监测系统中,一定空间范围内会部署多个按周期采样的传感器终端节点,这些节点的空间相对距离较近,

【共引文献】