复杂网络环境中网络流量分类的关键技术研究

The techniques of Internet traffic classification are the basis of applications for network management, traffic engineering, and network behavioral analysis, which divides the network traffic into multiple priorities or service classes. Internet traffic classification is a fundamental technology to mine useful information from the huge amount of Internet traffic data, thus supporting the security and management of network operations. However, existing work overlooks some important conditions and constraints of the Internet. In this project, we mainly research on the key techniques of Internet traffic classification in complex network. The project includes the following contents: 1. We study the techniques of real-time internet traffic classification in high-speed networks and construct a new online-classified method. 2. We explore the techniques of scalable traffic classification for mass of network traffic to solve the problem of concept drift. 3. We also study the techniques of robust classification, and design a new algorithm to classify the anomalous network traffic. 4. We explore the privacy problem, and put forward secure classification algorithms to protect user’s private information. By far, Internet traffic classification is a hot issue in both academic and industrial community. Therefore, this project is crucial not only in the theoretical search, but also in the real-world applications.

网络流量分类技术将复杂网络环境下的网络流量划分为多个优先级或多个服务类，因而是网络管理、流量工程、网络安全行为分析等网络研究应用的基础。网络流量分类技术可以帮助我们从复杂网络流量中去挖掘有用的信息，因此可以对网络运行进行有效的监管。然而，现有复杂网络环境下的网络流量分类方法还存在一定的问题。本项目主要研究复杂网络环境中网络流量分类的关键技术，具体包括：1. 研究高速网络环境下的实时流量分类技术，构造可在线分类网络流量的新方法；2. 研究海量网络流量的可扩展分类技术，解决现有网络流量分类算法的概念漂移和扩展性差等问题；3. 研究异常网络流量的鲁棒分类技术，设计可准确分类异常流量的新算法；4. 研究具有隐私保护的网络流量安全分类技术，提出保护用户隐私的新方案。目前，网络流量分类已成为学术界和产业界共同关注的研究热点。本项目的研究不仅具有重要的理论价值，而且具有巨大的实用价值。

互联网、云计算等网络技术的迅速发展大大拓展了网络连通范围，丰富了网络服务领域，使得网络空间已经成为人类活动新领域，人们对网络的依赖达到前所未有的程度。本项目从复杂网络环境和云环境中的关键问题入手，主要研究复杂网络环境下流量分类的关键问题，包括高速网络环境下的实时流量分类问题、海量网络流量下的分类问题、异常网络流量的鲁棒分类问题和具有隐私保护的网络流量安全分类问题。在项目执行期间，共发表论文8篇，其中SCI检索7篇，包括SCI一区期刊IEEE Internet of Things Journal和Information Sciences论文共计4篇，并获得发明专利授权2项。此外，项目负责人因在网络与系统安全领域取得的突出成就，当选为IEEE Fellow。项目负责人在项目执行期间协助举办国际会议7次，参加10余次国内外学术会议，培养博士生3名。项目负责人作为主编出版学术专著系列《SpringerBriefs on Cyber Security Systems and Networks》。项目执行情况良好，完成项目所有目标。