SIMULATION AND PERFORMANCE EVALUATION OF AN ATM MULTIPLEXER USING PRIORITY SCHEDULING

In an integrated services digital network with broadband capabilities (BISDN), constant bit rate (CBR) services as well as variable bit rate (VBR) services will ceexist. Services with different needs will be supported. To accommodate these diverse requirements, a connection-oriented packeblike switching and multiplexing principle called asynchronous transfer mode (ATM) has been proposed. In addition to the flexibility in supporting multi-rates, one benefit of the ATM approach is the potential of realizing better resource sharing in the BISDN environment. The objective of our study is to find some quantitative measurement that will help us better understand whether "sufficiently" more statistical gain can be realized with the added priority scheduling, i.e., by using delay priorities. A simulation model is developed to examine the delay performance and buffer occupancy statistics of an ATM multiplexer where the load generated from constant bit rate sources is held fiied, but the load generated by variable bit rate sources is varied. The performance of an ATM multiplexer that uses priority and one that uses no priority are compared. Notice that a wide spectrum of services are to be supported in B ISDN. Some services will tolerate less delay than others. Services may be separated into different delay priority classes based on their maximum tolerable delay requirements. By priority scheduling, one may serve first cells that can tolerate less delay, for example, those of a CBR service providing circuit emulation of a DSl or DS3. In this paper, we will investigate if, using priority scheduling coupled with a certain amount of buffering, the resource utilization efficiency can be improved and the service quality requirements for all services can still be met. A simulation model is developed to examine the delay performance and buffer occupancy statistics of an ATM multiplexer where the load ,generated from constant bit rate sources is held fiied, but the load generated by variable bit rate sources is varied. The objective of our study is to find some quantitative measurement that will help us better understand whether "sufficiently" more resource utilization gain can be realized with the added priority scheduling, i.e., by using delay priorities. The performance of an ATM multiplexer that uses priority and one that uses no priority are compared.