Residential Network Traffic and User Behavior Analysis

Internet usage is changing and the demands on the broadband networks are ever increasing. So it is still crucial to understand today's network traffic and the usage patterns of the end users, which will lead to more efficient network design, energy and costs savings, and improvement of the service offered to end users. This thesis aims at finding hidden patterns of traffic and user behavior in a residential fiber based access network. To address the problem, a systematic framework of traffic measurement and analysis is developed. It involves PacketLogic traffic data collecting, MySQL database storing, and traffic and user behavior analysis by using Python scripts. Our approach provides new insights on residential network traffic properties and Internet user habits of households, covering topics of aggregated traffic pattern, household traffic modeling, traffic and user penetration for applications, grouping analysis by cluster and subscriber, and concurrent application analysis. The analysis solutions we provide are based on open source tools without proprietary, giving the most flexibility for codes modification and distribution