Scalable fluid models and simulations for large-scale IP networks

In this article we present a scalable model of a network of Active Queue Management (AQM) routers serving a large population of Transport Control Protocol (TCP) flows. We present efficient solution techniques that allow one to obtain the transient behavior of the average queue lengths and packet loss/mark probabilities of AQM routers, and average end-to-end throughput and latencies of TCP users. We model different versions of TCP as well as different implementations of RED Random Early Detection (RED), the most popular AQM scheme currently in use. Comparisons between the models and ns simulation show our models to be quite accurate while at the same time requiring substantially less time to solve than packet level simulations, especially when workloads and bandwidths are high.