Fluid models and solutions for large-scale IP networks

Abstract

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

An Extension of the TCP Steady-State Throughput Equation for Parallel Flows and Its Application in MulTFRC

“…The congestion window size of the cumulative flow is given in (18), and the number of flows in the slow start phase in a TD-period is Using expressions for from (14) and for from (23), we have (24) Finally, the steady-state throughput of parallel flows can be calculated as (25) where is given in (24) and is defined in (17).…”

“…There is now a wealth of TCP models available (e.g., [11]- [14]), many of which are better than the model in [1] in some aspect, but they are all much more complex, making their practical application much harder. Some effort was also made to model a number of TCP flows sharing a bottleneck instead of focusing on just a single one [14]- [21]. The usability of these works is limited because they assume that the flows under consideration are alone in the network.…”

Variable Step Fluid Simulation for Communication Network

“…There has been much research on the network simulation based on fluid models [3,4,9,8]. Liu et al [3] compare packet-level simulations and fluid models in terms of relative efficiency.…”

“…A scalable model of a network of AQM routers is presented in [9], and the transient behavior of the average queue length, packet loss probabilities, and average end-to-end latencies have been observed. It is shown that their fluid model is accurate and requires substantially less time to solve, especially when workloads and bandwidths are high.…”

“…However, this model is based on the expected values of variables and the accuracy of results only applies when there exist large number of flows. A fluid model introduced in [9] also captures average window and queue sizes, but these statistics are not very accurate compared to the statistics of packet-level simulations.…”