Abstract
The Internet is a collection of more than 20,000 Autonomous Systems (ASes), each being an administrative region that employs its own network policies and protocols. Internet Service Providers (ISPs) are carrying increasing amounts of outbound traffic that flows through and out of their networks. To manage their networks, ISPs employ Traffic Engineering (TE) to predict and configure the traffic routing behavior so as to optimise IP network performance. TE can be classified into intra- and inter-AS. Intra-AS TE aims at controlling traffic routing only within the network while inter-AS TE aims to control traffic entering and exiting the AS. In this thesis, we investigate how to effectively apply both intra- and inter-AS TE to network dimensioning for ISPs, thereby assigning physical network resources to the forecasted traffic. There are three main contributions in this thesis: 1. Maximizing the network's ability to accommodate more future traffic demands. Due to their interaction, performing intra- and inter-AS TE separately may achieve sub-optimal network performance. We propose a joint TE optimization approach to allow ISP networks to accommodate significantly more future traffic demands. 2. Enhancing the robustness of traffic engineering solutions against traffic demand uncertainty. An accurate traffic matrix is an essential input to TE. However, due to dynamic network conditions and lack of perfect traffic measurement infrastracture, traffic demands are likely derived with uncertainty. We propose a scenario-based robust optimization approach to achieve more robust TE solutions against traffic demand uncertainty. 3. Achieving low-cost and resource-efficient end-to-end bandwidth guarantee provisioning: End-to-end bandwidth guarantees across ASes are vital for achieving Quality of Service (QoS) for newly emerging QoS-aware applications. ISPs have to determine the minimum bandwidth to be purchased from downstream ASes with as low cost as possible and then to efficiently assign routes towards destinations to customer traffic with bandwidth guarantees based on the accessible bandwidth. We propose a network dimensioning system that consists of novel problem formulations and efficient algorithms to achieve this end-to-end bandwidth guarantees provisioning. Key words: Traffic Engineering, Quality of Service, Robustness, Inter-AS, Intra-AS.