Analysis of the ARB (adaptive rate based) congestion avoidance algorithm

Artiges, Damien

doi:10.1109/cdc.1996.573561

Cited by 3 publications

(1 citation statement)

References 6 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…ARB belongs to this class of algorithm . Also in (Artiges, 1994), the author analytically proves the stability of the ARB algorithm based on a single source model.…”

Section: The Arb Algorithmmentioning

confidence: 97%

Analysis of An End-to-End Proportional Bandwidth Allocation Algorithm

Huynh

Nilsson

1997

High Performance Networking VII

View full text Add to dashboard Cite

In this paper, we present a new proportional bandwidth allocation scheme along with its analysis. The novel approach of this scheme is that it is purely end-to-end, it does not require any processing by the network, it is adaptive to changing network conditions, it has minimum oscillations, and it works even when sources have different end-to-end propagation delays. This scheme is a function of an Adaptive RateBased (ARB) congestion control algorithm that is designed to support best-effort traffic for which the proportional bandwidth allocation function is very desirable. We will briefly describe the ARB algorithm. We then analyze the behavior of the bottleneck queue, the settings of the ARB parameters and their effects on the behavior. This analysis will lead us to two conditions that. when they are satisfied , will provide proportional bandwidth allocation among the sources that are controlled by the ARB algorithm. We will show what parameters and how they can be controlled in order to achieve the proportional bandwidth allocation . Simulation will be used to validate our analysis. KeywordsAdaptive rate-based congestion control , proportional bandwidth allocation , performance analYSis A. Tantawy (ed.), High Performance Networking VII

show abstract

“…ARB belongs to this class of algorithm . Also in (Artiges, 1994), the author analytically proves the stability of the ARB algorithm based on a single source model.…”

Section: The Arb Algorithmmentioning

confidence: 97%