Effects of bursty and correlated traffic on the performance of Active Queue Management schemes

Motivated by the need of performance analysis of Active Queue Management (AQM) schemes under highly bursty and correlated traffic in multi-service networks, this paper develops an analytical model for a finite capacity queueing system with an AQM scheme subject to two heterogeneous classes of traffic which are modeled, respectively, by the bursty Markov-Modulated Poisson Process (MMPP) and non-bursty Poisson Process. The well-known MMPP is adopted for modelling bursty traffic owing to its ability to model the time-varying arrival rate and capture the important correlation between inter-arrival times while still maintaining analytical tractability. We derive the aggregate and marginal performance metrics including the mean queue length, response time, system utilization, throughput and loss probability. The accuracy of the model is validated by comparing analytical results with those obtained from simulation. We further focus on the evaluation of the effects of traffic burstiness and correlation of inter-arrival time on the aggregate performance metrics of AQM with different thresholds.