带野值处理的轻量级数据融合算法仿真研究

摘要

According to the promble of outlier, data loss widely existed in data set, energy and computation resources constraints in wireless sensor networks, a tiny aggregation algorithm with outlier preprocessing is proposed.It is designed to deal with outlier and aggregate data in local cluster of hierarchical clustering networks, based on Grubbs criteria and fuzzy weighted fusion algorithm.It' s simplicity and effectively to suit sensor nodes with limited resource.A real-world data set is used to verify our algorithm.The results show that it' s effective to eliminate outlier and make up impact of data loss.The accuracy of data sent to sink node are accordingly improved and energy is saved.%针对无线传感器网络节点感知数据存在野值、丢失以及能量和计算能力受限问题,提出了一种带野值预处理功能的轻量级数据融合算法.该算法基于Grubbs准则和模糊加权融合方法进行设计,主要用于剔除数据野值和实现多传感器数据融合.该算法设计简单、计算量小,能够满足节点有限计算能力的要求.采用第三方实测数据验证结果表明,该算法能有效地剔除感知数据集中的野值,弥补感知数据丢失带来的影响和提高传回汇聚节点数据的准确度.