基于能量感知决策树和增强boosting的网络数据聚合模型  

Research of Data Aggregation Model based on Energy Aware Decision Tree and Enhanced Boosting

作　　者：杜玉香[1] 陈欣 DU Yu-xiang;CHEN Xin(School of Inteligent Engineering,Guangzhou Nanyang Polytechnic College,Guangdong Guangzhou 510925,China;Department of Information Engineering,Zunyi Normal College,Guizhou Zunyi 563000,China)

机构地区：[1]广州南洋理工职业学院智能工程学院,广东广州510925 [2]遵义师范学院信息工程学院,贵州遵义563006

出　　处：《淮阴师范学院学报（自然科学版）》2022年第3期215-222,共8页Journal of Huaiyin Teachers College;Natural Science Edition

基　　金：广东省普通高校青年创新人才项目(2020KQNCX245)。

摘　　要：为了改善无线传感器网络(WSN)中的能量使用和数据传递,提出一种基于决策树桩和增强boosting的数据聚合模型.首先,计算出每个传感器节点的剩余能量,执行节点分类.然后,应用增强boosting提升分类效果,以增加不同类别的训练样本(传感器节点)之间的间隔,通过合并弱决策树桩的结果来构建强分类器,达到准确分类高、低能量节点的目的.最后,低能量传感器节点将数据包传递至邻近的高能量传感器节点,并通过汇聚节点采集数据包.实验从能耗、延迟、数据聚合准确度等方面进行评估,实验结果表明,与其他模型相比,所提模型的能耗和延迟,分别降低了26%和30%.数据聚合准确度和网络工作寿命提高了10%和9%.To improve energy use and data transmission in wireless sensor networks(WSN),a data aggregation model based on decision tree stump and linear programming is proposed.Firstly,the residual energy of each sensor node is calculated and node classification is performed.Then,enhanced boosting model is applied to increase the interval between different types of training samples(sensor nodes),and a strong classifier is constructed by combining the results of weak decision tree stumps to accurately classify high and low-energy nodes.Finally,the low-energy sensor node transmits the data packet to the adjacent high-energy sensor node,and collects the data packet through the sink node.The experimental results show that compared with other excellent models,the energy consumption and delay of the proposed model are reduced by 22% and 35%.Data aggregation accuracy and network lifetime increased by at least 13%and 10%.

关 键 词：无线传感器网络 决策树桩 增强boosting 传感器节点 分类器 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]