摘要
P2P流媒体流量中的控制流与数据流,由于统计特征差异较大,致使DFI(深度流检测)方法识别其效果不佳。借鉴DFI的思想,提出一种基于端点特征识别P2P流媒体流量的方法。该方法针对网络端点,提取了六个有效特征,并结合机器学习的方法识别P2P流媒体流量。实验结果表明,该方法比DFI识别的整体准确率要高,且可以用于P2P流媒体的在线识别。
Due to the great differences in statistical features between control flows and data flows,the performance of DFI（deep flow inspection） in the identification of P2P streaming traffic is not so ideal.Enlightened by the idea of DFI,this paper proposed a P2P streaming identification method based on endpoint features.This method chose six features aimed at net-endpoint so as to identify P2P streaming traffic using machine learning.Experimental results show that this method performs better in overall accuracy over DFI,and it can also be used in real-time P2P streaming traffic identification.
出处
《计算机应用研究》
CSCD
北大核心
2012年第7期2600-2602,2606,共4页
Application Research of Computers
基金
国家科技支撑计划课题(2011BAH19B01)
关键词
P2P流媒体
机器学习
流
端点
P2P streaming
machine learning
flow
endpoint
