Analytical modeling for the threshold service differentiation mechanism in asynchronous optical buffers

Abstract

With the emergence of different kinds of applications, service differentiation has become an important issue to be considered in current and future networks. We propose an exact analytical model for evaluating the performance of feed-forward delay-line buffers when the threshold mechanism is used for service differentiation in asynchronous optical packet-switched networks. The analytical model was derived assuming Poisson arrivals and any packet length distribution. Parameters such as buffer size, load, and others do not affect the accuracy of the model. In addition, it can use an arbitrary number of service-differentiated classes and traffic partitioning within them. To check the exactness of the model, we compared the buffer modeling with simulation results when exponential and uniform distributions are considered for the packet length. The analysis presented here shows that by using the threshold mechanism, it is possible to effectively differentiate the per class packet blocking probability.