Comparative analysis of behavioral structure from internet traffic

Understanding the characteristics of Internet application structure and traffic is critical for traffic modeling and network optimization. In this paper, the behavioral structures of different Internet applications are constructed based on a large sample of flow data collected from operational network of ISP. For each application type, i.e. Web, P2P Stream, P2P Download, Video Stream, Instant Messaging (IM), FTP, Email, and VoIP, the behavioral structure is modeled as complex network. The statistical characteristics of these complex networks, including the degree distribution of hosts, the distance between hosts and community structure, are investigated and compared. Comparative analysis results show that the behavioral structures of Internet applications exhibit characteristics such as scale-free, small-world and highly clustered, which imply the heterogeneous behavior of users on the Internet.