THESIS
1996
1, vi, 69 leaves : ill. ; 30 cm
Abstract
Performance analysis of ATM switches has been an active research topic in these recent years. Numerous models have been proposed throughout the world. Most of these models are developed for analyzing specific ATM switches with specific kinds of incoming traffic. The resultant models are usually highly dependent on the structure of the switches. This restrict the usage of these models to other switches with different structures. In addition most of these models are built with two unrealistic assumptions: the incoming traffic has random destinations and the size of the switch tends to infinity....[
Read more ]
Performance analysis of ATM switches has been an active research topic in these recent years. Numerous models have been proposed throughout the world. Most of these models are developed for analyzing specific ATM switches with specific kinds of incoming traffic. The resultant models are usually highly dependent on the structure of the switches. This restrict the usage of these models to other switches with different structures. In addition most of these models are built with two unrealistic assumptions: the incoming traffic has random destinations and the size of the switch tends to infinity.
In this research an analytical model for the performance analysis of ATM switches is developed to overcome the above problems. This model can be applied to a large class of ATM switches called the input-output buffering ATM switches. We consider general incoming traffic with asymmetrical output loading and correlated destinations which is most common in ATM networks. Our model is constructed by studying the behavior of the switch, i.e., blocking in the switch fabric and the correlation of incoming traffic. The switch fabric is then modeled as a closed queueing network. By changing the characteristics of this closed queueing network, different types of switches can be studied. This model gives accurate results for the maximum throughput, the cell loss probability and the mean response time under different traffic conditions.
Post a Comment