An evaluation framework for active queue management schemes

Abstract

Over the last decade numerous active queue management (AQM) schemes have been proposed in the literature. Many of these studies have been directed towards improving congestion control in best-effort networks. However, there has been a notable lack of standardised performance evaluation of AQM schemes. A rigorous study of the influence of parameterisation on specific schemes and the establishment of common comparison criteria is essential for objective evaluation of the different approaches. A framework for the detailed evaluation of AQM schemes is described in this paper. This provides a deceptively simple user interface whilst maximally exploiting relevant features of the NS2 simulator. Traffic models and network topologies are carefully chosen to characterise the target simulation environment. The credibility of the results obtained is enhanced by vigilant treatment of the simulation data. The impact of AQM schemes on global network performance is assessed using five carefully selected metrics. Thus, a comprehensive evaluation of AQM schemes may be achieved using the proposed framework.