基于深度学习的通信网关键节点自动识别方法  

Automatic Identification Method of Key Nodes in Communication Network Based on Deep Learning

作　　者：刘建设 LIU Jian-she(Hebei Civil Air Defense 218 Project Management Station,Shijiazhuang 050200 China)

机构地区：[1]河北省人防218工程管理站,河北石家庄050200

出　　处：《自动化技术与应用》2024年第3期104-107,共4页Techniques of Automation and Applications

摘　　要：短波通信网关键节点具有直接影响网络安全及通信稳定的能力,为此,提出基于深度学习的通信网关键节点自动识别方法。建立通信节点重要度水平评价体系,从三个方面分析节点重要度,找出重要度较高节点;构建多层卷积神经网络,降维处理输入节点样本,经过卷积和池化作用,得到节点特征信息;计算每个节点样本与某个类别间所属概率值,其中概率值最大类别即为短波通信网关键节点。以某地短波通信网展开仿真测试,结果表明,所提方法具有理想的关键节点识别精度,为加强关键节点保护力度、避免网络崩溃事件发生提供了有效的数据基础。Key nodes in HF communication network have the ability to directly affect network security and communication stability.Therefore,an automatic identification method of key nodes in communication network based on deep learning is proposed.It establishes the evaluation system of communication node importance level,analyzes the node importance from three aspects,and find out the nodes with high importance.The multi-layer convolution neural network is constructed,the dimension of the input node samples is reduced,and the node characteristic information is obtained through convolution and pooling.It calculates the probability value between each node sample and a category,in which the category with the largest probability value is the key node of HF communication network.The simulation results of a short wave communication network show that the proposed method has ideal key node identification accuracy,and provides an effective data basis for strengthening the protection of key nodes and avoiding network collapse.

关 键 词：卷积神经网络 短波通信网 节点重要度 关键节点识别 拓扑位置 池化作用 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]