Research project area: Interdomain and Overlay Networks

Responsible: Řivind Kure

Abstract: The goal of the project is to investigate how available QoS elements in the unicast domain can be utilized by overlay networks and interdomain networks. QoS has to a large extent been investigated in an intra-domain setting, but there has been less focus on various overlay networks. There are many different types of overlay networks. The ones we will be focusing on are multiparty distribution networks, mobility networks, ad-hoc networks, and IP-sec overlay networks. These networks will be built on top of unicast intradomain networks with call admission, dynamic adaptation of QoS on path, and robustness. These elements or mechanisms need to be adapted to the overlay network, and even in some cases be replaced by different mechanisms. In particular we will target measurement based call admission, handling of dynamic QoS, and dynamic rerouting.

Compared to the intradomain setting, there has been a limited effort within interdomain QoS, and without an apparent success. A particular lacking area is experimental verification of various ideas. The aim will be limited to usability of measurements for interdomain QoS routing. Establishing an experimental basis is clearly outside the scope of the project. However, as part of the overall problems setting, we shall establish a simulation platform that can be used by another more experimental investigation in the interdomain QoS issue.

Important problems are:

• Suitability of existing SLA proposals for overlay networks
• Dynamic QoS paths in overlay networks and rerouting strategies
• Measurement methodology for overlay networks
• Traffic characterization
• Interaction between QoS mechanisms at the intradomain level, routing and load balancing techniques in this class of networks
• Measurement based admission control in overlay networks
• Interaction with security mechanisms