Research project area: Intradomain Networks

Responsible: Peder J. Emstad

Abstract: The networks are designed and managed by the same entity, which enables a coordinated use of resources within the network to provide the appropriate QoS. This project will deal with how such coordination should be performed, its quantitative effects on the QoS as well as underlying issues such as offered traffic and perceived QoS. The cooperation with UNINETT allows large-scale experiments with respect to this kind of problems. The networks may be of different types, such as mobile, wireless and fixed networks.

The 3rd generation of mobile networks will use IP networks for all services. Label switching is a new attractive technique that may be used also in optical networks. The intent to use label switching (GMPLS) as a unified means to control network resources in several of todays network layers, represents a new range of challenges. While capacity in principle can be made available in fixed networks, it will remain a scarce resource in mobile and wireless networks where efficient spectrum usage and channel access methods are important. The network nodes will typically carry high volumes of traffic, and end user QoS- requirements will force differentiated handling of the traffic. User mobility in a wide sense, abnormal user behaviour, occurrence of failures and the known variance in IP-traffic represents further challenges and prohibits over-provisioning as a general solution.

Important problems are:

• Characterization of offered load and user expectation
• Resource management (priorities, reservation, policies) and SLAs
• Traffic engineering, routing, and load balancing
• Fault management, network survivability and QoS differentiation
• Admission control
• Interaction between application/terminal equipment and network
• Laboratory experiments with live traffic over UNINETT and studies of QoS
• Experiments with management and control mechanisms in laboratory environment or in the UNINETT research network
• Measurement techniques and statistical problems
• Finding relations between perceived QoS and network measures such as delay and loss
• Dynamic mechanisms for handling QoS in heterogenious mobile environments