Resource Management and Traffic Control in Large Broadband Networks under QoS ConstraintsNational projects
January 15, 2000 - July 14, 2001
NCSR budget: 48,000,000€
The project employs asymptotic techniques to develop a theory that will allow, in simple but sufficiently accurate terms, the quantitative characterisation of the economies of scale achievable in large multiplexing environments, reminiscent of the operating conditions in modern large, high-speed backbones. The resulting theory will provide the basis for the design of enhanced procedures for (a) network resource allocation, under given traffic load and QoS constraints and (b) connection admission control, under given resources and QoS requirements. In the studied regime, a simple traffic characterization through just second order statistical quantities (which are relatively easy to compute or measure directly) will suffice.The theoretical results will be verified through experimentation, during which an ATM switch will be congested under controlled traffic load, provided by parameterisable traffic generators. The data losses will be traced and the outcome of the experiments will be compared against the theoretical predictions. A primary objective of the experiments is to validate the assumptions embedded in the asymptotic regime providing the foundation of the theory.
Finally, the project intends to provide specifications allowing the transparent integration of the new traffic control and resource management schemes into broadband network management platforms. The relevant study will focus on distributed object-oriented architectures and will employ the notions of abstract information models and intelligent software agents.