基于互联网流量识别对于互联网的管理与安全的重要性,就互联网流量识别产生的背景和主要的识别方法进行全面的阐述。针对基于机器学习的流量识别、早期流量识别以及实时在线流量识别3个研究热点,阐述了国内外的研究进展情况。总结了当前互联网流量识别研究中存在的关键技术问题,包括基础数据采集困难、特征提取技术以及非平衡识别技术有待突破等问题。指出该领域研究未来发展的3个重要方向:移动互联网的流量识别;高速网络环境下的流量识别;云计算环境下的流量识别与分析。
Based on the importance of internet traffic identification for the management and security of internet,we firstly describe the background and main techniques,and then discuss the research circumstances of internet traffic identification with regard to machine learning based traffic identification,early stage traffic identification and real-time or online traffic identification. The existing key technical issues of internet traffic identification,such as basic data collection,feature extraction and imbalanced traffic data identification,are summarized. Finally three future research directions,including traffic identification of mobile Internet,traffic identification for high speed networks and identification and analysis of cloud computing traffic,are proposed.