基于SDN的一体化信息网络业务流分类策略  

Traffic classification strategy of integrated information network based on SDN

作　　者：赵曦 马礼[1] 傅颖勋 李阳 马东超[1] ZHAO Xi;MA Li;FU Ying-xun;LI Yang;MA Dong-chao(School of Information Science and Technology,North China University of Technology,Beijing 100144,China)

机构地区：[1]北方工业大学信息学院,北京100144

出　　处：《计算机工程与设计》2023年第1期30-36,共7页Computer Engineering and Design

基　　金：国家重点研发计划基金项目(2018YFB1800302);国家自然科学基金项目(62001007);北京市自然科学基金项目(KZ201810009011、4202020、19L2021)。

摘　　要：一体化信息网络有多种类型协议的网络接入,流量管理与调度困难,针对这种情况设计基于软件定义网络(SDN)架构的一体化信息网络业务流分类技术。根据SDN框架转控分离的特点,设计流量感知节点采集流量信息,针对一体化信息网络流量连续特征属性多、业务类别分布不平衡、存在大量噪声的特点,设计基于Fayyad边界点定理改进CART算法,与基于弱分类器系数和分类误差相似度改进的Adaboost算法相结合的分类模型。实验结果表明,该技术能够对采集到的业务流进行分类,与相关算法相比,分类精度和用时均有明显改进。The integrated information network has network access of various types of protocols,and traffic management and scheduling are difficult.In view of this situation,a service flow classification technology of integrated information network based on software defined network(SDN)architecture was designed.According to the characteristics of transfer and control separation of SDN framework,a traffic sensing node was designed to collect traffic information.According to the characteristics of continuous traffic characteristic attributes,unbalanced service category distribution and large amount of noise in the integrated information network,an improved CART algorithm based on Fayyad boundary point theorem,and the classification model was combined with the improved AdaBoost algorithm based on weak classifier coefficient and classification error similarity was designed.Experimental results show that this technology can classify the collected traffic flow,and the classification accuracy and time are significantly improved compared with the relevant algorithms.

关 键 词：一体化信息网络 SDN(软件定义网络) 业务流分类 CART(分类与回归树) Adaboost(自适应提升算法) Fayyad边界点定理 集成学习 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]