集成分类器的加密YouTube流量精细化分类方法  

An Integrated Classifier to Encrypt YouTube Traffic Fine Classification Method

作　　者：邹乐述 翟江涛 ZOU Leshu;ZHAI Jiangtao(College of Electronics&Information,Jiangsu University of Science&Technology,Zhenjiang 212003,China;College of Electronics&Information Engineering,Nanjing University of Information Science&Technology,Nanjing 210044,China)

机构地区：[1]江苏科技大学电子信息学院,江苏镇江212003 [2]南京信息工程大学电子与信息工程学院,南京210044

出　　处：《重庆理工大学学报（自然科学）》2021年第6期165-173,共9页Journal of Chongqing University of Technology：Natural Science

基　　金：国家自然科学基金项目(61702235)。

摘　　要：为提升网络服务质量,实现流量精细化可管可控,针对特征失效问题,提出一种加密YouTube视频流量的精细化分类方法。设计快捷有效的特征提取方法,同时,为解决机器学习单个分类器度量手段单一问题,选取不同种类分类器,经过特征筛选后为每个分类器输入不同的特征数量组合。通过设置权值和阈值,根据分类精度进行权值更新,最终实现高精度分类。实验结果表明:所提方法较现有模型对加密应用下流量识别效果提升3%左右。Application classification of encrypted traffic is one of the hot topics in current research.However,the current algorithm has the problem that classification particles are too thick to meet the current needs.In order to improve the service quality of the network and realize the traffic fine and controllable,a fine classification method of encrypting YouTube video traffic is proposed for the feature failure problem.First,a fast and effective feature extraction method is designed.At the same time,in order to solve the single measurement method of a single classifier in machine learning,different types of classifiers are selected and different number combinations of features are input for each classifier after feature screening.By setting the weights and thresholds,the weights are updated according to the classification accuracy,and the classification with high precision is finally realized.The experimental results show that the proposed method is about 3%more effective than the existing model in traffic recognition.

关 键 词：YouTube视频流量 特征提取 集成学习 权值更新 流量分类 

分 类 号：TP393.08[自动化与计算机技术—计算机应用技术]