Rate-based versus queue-based models of congestion control

Abstract

Mathematical models of congestion control capture the congestion indication mechanism at the router in two different ways: rate-based models, where the queue-length at the router does not explicitly appear in the model, and queue-based models, where the queue length at the router is explicitly a part of the model. Even though most congestion indication mechanisms use the queue length to compute the packet marking or dropping probability to indicate congestion, we argue that, depending upon the choice of the parameters of the AQM scheme, one would obtain a rate-based model or a rate-and-queue-based model as the deterministic limit of a stochastic system with a large number of users. We also consider the impact of implementing AQM schemes in the real queue or a virtual queue. If an AQM scheme is implemented in a real queue, we show that, to ensure that the queuing delays are negligible compared to RTTs, one is forced to choose the parameters of a AQM scheme in a manner which yields a rate-based deterministic model. On the other hand, if the AQM scheme is implemented in a virtual queue, small-queue operation is achieved independent of the choice of the parameters, thus showing a robustness property of virtual queue-based schemes.