The rapid growth of Internet Traffic has emerged as a major issue due to the rapid development of various network applications and Internet services. One of the challenges facing Internet Service Providers (ISPs) is to optimize the performance of their networks in the face of continuously increasing amounts of IP traffic while guaranteeing some specific Quality of Services (QoS). Therefore it is necessary for ISPs to study the traffic patterns and user behaviors in different localities, to estimate the application usage trends, and thereby to come up with solutions that can effectively, efficiently, and economically support their users’ traffic.

The main objective of this thesis is to analyze and characterize traffic in a local multi-service residential IP network in Sweden (referred to in this report as “Network North”). The data about the amount of traffic was measured using a real-time traffic-monitoring tool from Packet Logic. Traffic from the monitored network to various destinations was captured and classified into 5 ring-wise locality levels in accordance with the traffic’s geographic destinations: traffic within Network North and traffic to the remainder of the North of Sweden, Sweden, Europe, and World.

Parameters such as traffic patterns (e.g., traffic volume distribution, application usage, and application popularity) and user behavior (e.g., usage habits, user interests, etc.) at different geographic localities were studied in this project. As a result of a

systematic and in-depth measurement and the fact that the number of content servers at the World, Europe, and Sweden levels are quite large, we recommend that an intelligent content distribution system be positioned at Level 1 localities in order to reduce the amount of duplicate traffic in the network and thereby removing this traffic load from the core network.

The results of these measurements provide a temporal reference for ISPs of their present traffic and should allow them to better manage their network. However, due to certain circumstances the analysis was limited due to the set of available daily traffic traces. To provide a more trustworthy solution, a relatively longer-term, periodic, and seasonal traffic analysis could be done in the future based on the established measurement framework.