Mathematical Modeling of the RED-RWM Active Queue Management Scheme

摘要

In this paper, we extend the work done on stochastic modelling of Random Early Detection with Explicit Congestion Control (RED-ECN) gateways to present a mathematical model for Random Early Detection with Receiver Window Modification (RED-RWM) gateways, a novel TCP congestion control approach, which could be used with any of the present Active Queue Management (AQM) schemes, such as Random Early Detection (RED), Adaptive Random Early Detection (ARED), and BLUE queues to further reduce congestion at the ingress and gateway routers. We also provide an asymptotic and steady-state analysis for the model and show that 1) the average TCP dynamics for a large number of flows using a RED-RWM queue is closely related to that of a single flow utilizing the same TCP congestion control mechanism and that 2) the sessions become asymptotically independent as number of flows using the queue becomes large, suggesting that the RED-RWM queue alleviates the synchronization problem among the flows. We use a Monte Carlo simulation method to compare outcomes of our mathematical model with outcomes of NS2 simulation experiments. In earlier published works, we had shown that RWM schemes complementing RED, Blue and ARED queues easily outperform RED, Blue and ARED queues functioning with and without ECN. Hence, by using the mathematical model, one can predict the outcome of the scheme before implementing it.