Abstract:
The process of traffic classification provides a substantial amount of information about the traffic activities in a network. It enables network operators and researchers to observe trends that can help them improve the network experience and services for their users. In this thesis, we investigate how the University of Auckland's Internet traffic has changed over a period of approximately ten years since 2006. We began by comparing and evaluating various traffic classifiers to find one that is most appropriate for our University dataset. We then used our chosen network traffic classifier to classify Internet traffic flows into classes that identify the applications responsible for producing them. Finally, we performed a detailed analysis on these classified application flows to observe for trends that help answer our research questions. The outcomes obtained from our research show that there are significant changes in the University's network traffic activity across the years. As an example, we found that network traffic activity at the University has shifted from being predominantly active in the morning to one that is active both in the morning and afternoon.