Traffic Control in a Synchronous Transfer Mode Networks
Khalil, Ibrahim (1995) Traffic Control in a Synchronous Transfer Mode Networks. Masters thesis, Universiti Pertanian Malaysia.
In the 90s, there is an increasing demand for new telecommunication services such as video conferencing, videophone, broadcast television, image transfer and bulk file transfe r etc. At the same time, transmission systems at bit rates of 2.5 Gb/s are now being installed, and the expected next generation of 10 Gb/s systems is emerging from the research laboratories. Coupled with that the development and deployment of new technologies systems such as fiber optics and intelligent high-speed switches have made it possible to provide these services in future high-speed integrated services networks like Asynchronous Transfer Mode (ATM). However, because of their new characteristics, these new services pose great challenges not previously encountered in traditional circuitswitche d or packet switched networks. For example, feature s such as large propagation delay as compared to transmission delay, diverse application demands, constraints on call processing capacity, and Quality-Of-Service (QOS) support for different applications all present new challenges arising from the new technology and new applications. Thus, much research is needed not just to improve existing technologies, but to seek a fundamentally different approach toward network architectures and protocols. In particular, new bandwidth allocation and call admission control algorithms need to be studied to meet these new challenges. A VP bandwidth allocation problem is studied for services which requires guaranteed connection for a fixed duration of time leading to extensive use of facilities like reservations of transmission capacity in advance. In such a case, the network may offer discounts for users reserving capacities in advance due to the advantage of working with predetermined traffic loads. Similarly, charges may differ for customers wanting to book capacity for a specified tie interval. Based on this scenario, various charge classes and booking policies are introduced. An effective bandwidth allocation scheme is proposed at the VP level with multiple nested charge classes where these various classes are allocated bandwidth optimally through some booking policies'. The scheme is also shown to be effective in maximizing network revenue. The best tradeoff between revenue gained through greater demand for discount bandwidth units against revenue lost when full-charge bookings request must be turned away because of prior bookings of discount bandwidth units is also sought for.
Repository Staff Only: Edit item detail