Stochastic reward net end-to-end quality of service (QoS) simulation modeling across ATM network: using the goodput model

Multiservice network architecture designers the world over are adopting asynchronous transmission mode (ATM) as the backbone data transmission protocol, because of ATM's tested-ability to support very high data rates in a multi-protocol and multilayered data network. To analyse and evaluate a multiservice network's overall performance, the effective throughput model has been used extensively. We employ the concept of "goodput" which describes more stringent parameters in modeling the end-to-end QoS of a multiservice network built on an ATM network. We have also derived a mathematical model for goodput and simulated the traffic situation using the stochastic reward net model. Performance indices, such as system goodput and cell loss ratio for priority cells and tagged cells, were measured. The simulation model was subjected to different buffer size (20, 40, 60) ATM cells, while varying the ratios of probabilities of priority cells to tagged cells (0.5:0.5, 0.6:0.4, 0.7:0.3). The network performance simulation showed a system goodput measure of approximately 97%. It was also observed that priority cell loss ratios fluctuated between 0.0014 and 0.0016 which is crucial for the deployment of a rate- and delay-sensitive data communication system.