Simple models for statistically multiplexed data traffic in cell relay networks

A straightforward method is presented for computing the performance of a queue with constant service rate and batch Poisson arrivals, where each batch (burst) consists of multiple cells. Closed-form formulas are derived for the queue length distribution, cell loss probability and delay distribution. The study applies to queues with both finite and infinite buffer size. We also reveal that the cell delay distribution is the same as the burst delay distribution in this queueing system. In addition, we illustrate that the formulas derived (e.g., cell loss probability) for the cell performance in a queue with batch Poisson arrivals and a constant service rate are upper bounds for the queueing performance of multiplexed on-off sources. The bounds become tight when the ratio of the queue service rate and on-off source peak rate is small