基于深度学习的SDN流量分类方法研究  

Research on SDN Traffic Classification Method Based on Deep Learning

作　　者：池亚平[1] 刘怡龙 许盛伟[1] CHI Yaping;LIU Yilong;XU Shengwei(Beijing Electronic Science and Technology Institute,Bejing 100070,P.R.China;Xidian University,Xi'an 710071,P.R.China)

机构地区：[1]北京电子科技学院,北京市100070 [2]西安电子科技大学,西安市710071

出　　处：《北京电子科技学院学报》2022年第2期112-121,共10页Journal of Beijing Electronic Science And Technology Institute

摘　　要：针对在SDN网络架构下数据流量在线分类问题,提出了一种基于格拉姆角场和卷积神经网络的流量分类方法。该方法利用SDN网络架构收集数据流的前几个数据包的少量统计特征,通过格拉姆角场实现特征的扩展,利用卷积神经网络实现数据流量的智能分类。通过与现有的分类方法对比分析,表明在获取的统计特征较少的情况下,本文提出的方法在分类准确率等方面表现更优。To address the problem of online data traffic classification under SDN network architecture,a traffic classification method based on the Gramian Angular field and the convolutional neural network is proposed.In this method,SDN network architecture is utilized to collect a small quantity of statistical characteristics of the first several data packets of the data flow.Expansion of the characteristics is realized by using the Gramian Angular Field,and the convolutional neural network is used to realize the intelligent classification of the data flow.A comparative analysis between the proposed method and existing classification methods indicates that the proposed method could achieve better classification accuracy in the case of few statistical features.

关 键 词：流量分类 SDN网络 格拉姆角场 卷积神经网络 

分 类 号：TN309[电子电信—物理电子学]