Research On Traffic Classification Mrthod And Traffic Characteristics Analysis

In recent years, Internet content services and infrastructure are developing fast, which then leads to the development of the Internet broadband applications. According to the calculation of China Internet Network Information Center, by the end of June2013, the number of Chinese Internet users has reached a total of591million. Internet penetration rate was44.1%, improving by3.4percentage points over last year. Internet applications are the root causes of the network traffic and network behavior. Accurately identifying these applications and analyzing the characteristics of network traffic is precondition for in-depth understanding of network conditions, user behavior and Internet status quo, which has important significance to not only the network administrator or service provider but also the users and researchers.Network traffic classification techniques, is one of the most basic and important means of analyzing the network traffic characteristics and enhancing controllability of network, which is widely used in network traffic monitoring, network security, billing management, user behavior analysis and other network activities. When analyzing characteristics of the fixed Internet, the rapid development of mobile Internet in recent years makes the analyzing of the mobile Internet traffic characteristics becoming another particularly important research. With more users participated in the Internet and higer line speed provided, the online business equipment every day can produce terabytes or even petabytes of traffic data. To storing and analyzing such mass data has become an important research for network traffic characteristics analysis.In order to analyzing the network traffic, we developed Traffic Analysis and Classification System (TACS) and big data analyzing system called LogAnalyser. This thesis introduced the basic method of network traffic classification and analyszing, after a brief introduction to TACS system, a deep research on classifier evaluation measure in network traffic classification was described in detail. On this basis, we conducted experiments about traffic classification based on Deep Flow Inspection and improved evaluation measure on the real network traffic flow. For analyzing massive network traffic data, our self-developed analysis system was introduced.Finally, on the basis of TACS and LogAnalyser, the traffic flow characteristics of fixed and mobile Internet network were analyzed from a different perspective.