基于多共引特征的ZigBee无线传感网络节点分类  

ZigBee Wireless Sensor Network Node Classification Based on Multiple Co-Citation Features

作　　者：周澄[1] 许胜[1] ZHOU Cheng;XU Sheng(School of Mechanical and Electrical Engineering,Taizhou University,Taizhou Jiangsu 225300,China)

机构地区：[1]泰州学院机电工程学院,江苏泰州225300

出　　处：《传感技术学报》2024年第11期1976-1982,共7页Chinese Journal of Sensors and Actuators

基　　金：江苏省高等学校自然科学研究面上项目(20KJB520016)。

摘　　要：ZigBee无线传感网络节点数量大、密度高、特征向量复杂,导致节点分类难度上升。在复杂的大差异节点特征中,研究以融合共引节点特征为基础的ZigBee无线传感网络节点分类方法。采集ZigBee无线传感网络节点数据,提取节点的embedding、入度、出度、集聚系数、PageRank值等有共引属性的特征,确定节点共引特征向量。利用搭建的节点分类模型,将提取的共引特征嵌入到模型空间内,通过引入注意力机制及多角度信息的融合完成分类模型的训练,利用训练好的模型对于ZigBee无线传感网络节点进行分类。实验结果表明,所提方法的micro-F1和macro-F1均随着节点的增多而降低,约为95.8、96.3,且并未出现上升的趋势,表明ZigBee无线传感网络节点分类效果好,综合效率高,可靠性强。The ZigBee wireless sensor network has a large number of nodes,it is of high density,and has complex feature vectors,which increases the difficulty of node classification.On the basis of complex and high differentiated node features,this paper studies the Zig-Bee wireless sensor network node classification method based on the fusion of co-referenced node features is studied.Data are collected from ZigBee wireless sensor network nodes,features with co-citation attributes are extracted such as embedding,in degree,out degree,clustering coefficient,PageRank value,etc.of nodes,and the co-citation feature vector of nodes is determined.By using the constructed node classification model,the extracted co-citation features are embedded into the model space.The training of the classification model is completed by introducing attention mechanisms and integrating multi angle information.The trained model is used to classify ZigBee wireless sensor network nodes.The experimental results show that both micro-F1 and macro-F1 of this method decrease with the increase of nodes,about 95.8 and 96.3,and there is no upward trend,indicating that ZigBee wireless sensor network has good node classification performance,high comprehensive efficiency,and strong reliability.

关 键 词：ZIGBEE无线传感网络 节点分类 共引特征 特征提取 特征向量 分类训练 注意力机制 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] TP311[自动化与计算机技术—控制科学与工程]