基于多特征的P2P直播流识别方法  

Method of Identifying P2P Live Streaming Based on Union Features

作　　者：杨楷[1] 汪斌强[2] 王浩学[2] 

机构地区：[1]国家数字交换系统工程技术研究中心,河南郑州450002 [2]信息工程大学科研部,河南郑州450001

出　　处：《信息工程大学学报》2014年第1期98-104,共7页Journal of Information Engineering University

基　　金：国家863计划资助项目(2011AA01A103)

摘　　要：基于P2P的直播应用是当前发展最为迅速的因特网应用之一,实时识别P2P直播流是管理网络P2P流媒体流量和理解网络行为的关键一步。针对当前P2P流媒体直播流的识别方法较少,识别效果一般,文章分析了P2P直播流的行为特征,提出了基于节点连接度的识别方法和基于BM信息比的识别方法,并结合两个流量特征采用联合特征进行P2P直播识别。实验表明该方法识别整体准确率较高,可以实现P2P直播的在线识别。P2P live is one of the rapidly developing internet applications. It is the key step for network management identifying P2P live streaming accurately and on time, but currently only a few papers focusing on identifying P2P live streaming. This paper analyzes the behavioral characteristics of P2P live streaming, and proposes an identification method based on node connectivity or BM packet. Then it combines two flow characteristics for identifying P2P live streaming. The experiment shows that the recognition method based on union features can accurately and quickly identify P2P live streaming.

关 键 词：P2P直播 连接度 BM包 联合特征 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]