Priority queuing of long-range dependent traffic

This paper presents an analytical framework to estimate the queue length and delay survivor functions for a priority queuing system with varying service rate. The approach is based on multi-scale queuing (MSQ) and assumes that both the traffic and service rate are characterized using multi-scale multiplicative models. To evaluate the performance of a given queue, the paper models the multi-queue as a two-queue system. The two-queue system aggregates all traffic of priority higher than the queue under consideration into one higher-priority composite queue. The queue under consideration is then analyzed as the lower priority queue in this simpler two-queue system. To enable the analysis, the paper first introduces a technique to estimate the multi-scale model for an aggregate of traffic sources. The paper then provides means for using variable service multi-scale queuing (VS-MSQ) to obtain the multi-scale characteristics of the remaining capacity available to the lower-priority queue under consideration, and subsequently, to estimate the lower-priority queue length, and delay distributions using VS-MSQ. Comparison with simulation results shows that the analytical results can provide accurate estimation of the queue length and delay distributions.