软件定义网络面向异常流提取的自适应流抽样算法  

An Adaptive Flow Sampling Algorithm for Abnormal Flow Extraction in Software Defined Networks

作　　者：谢欣 XIE Xin(Faculty of Information Engineering and Automation, Kunming University of Science and Technology;Yunnan Key Laboratory of Artificial Intelligence, Kunming University of Science and Technology;Key Laboratory of Computer Technology Application of Yunnan Province, Kunming University of Science and Technology)

机构地区：[1]昆明理工大学信息工程与自动化学院 [2]昆明理工大学云南省人工智能重点实验室 [3]昆明理工大学云南省计算机技术应用重点实验室

出　　处：《化工自动化及仪表》2022年第5期611-618,共8页Control and Instruments in Chemical Industry

基　　金：2020年云南省研究生优质课程“算法分析与设计”建设项目(109920210048);云南省“万人计划”产业技术领军人才项目(云发改人事[2019]1096号);云南省技术创新人才项目(2019HB113)。

摘　　要：针对现有传统网络流量抽样技术难以应对动态多变的高维度流量数据,造成大量网络资源开销且无法保证小流的抽样精度,对异常流感知能力较弱的问题,提出一种基于软件定义网络面向异常流提取的自适应流抽样算法。经仿真实验佐证,该算法能在确保抽样精度的同时减轻网络负载与通信开销,对异常流提取能力显著提升。The existing network traffic sampling technology has difficulty in dealing with dynamic and changeable high-dimensional traffic data and this may cause a lot of network resource overhead and fail to guarantee the sampling accuracy of small flows and the ability to perceive abnormal flows is weak.In this paper,an adaptive flow sampling algorithm based on software defined network for anomalous flow extraction was proposed.Simulation results show that,the proposed algorithm can reduce the network load and communication overhead while ensuring the sampling accuracy,and significantly improve the ability of abnormal flow extraction.

关 键 词：软件定义网络 自适应流抽样 动态轮询 介数中心性 近似曲率 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]