Convergence complexity of optimistic rate based flow control algorithms

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

This paper studies basic properties of rate based flowcontrol algorithms and of the max-min fairness criteria. For the algorithms we suggest a new approach for their modeling and analysis, which may be considered more "optimistic" and realistic than traditional approaches. Three variations of the approach are presented and their rate of convergence to an optimal max-min fairness solution is analyzed. In addition, we introduce and analyze approximate rate based flow control algorithms. We show that under certain conditions the approximate algorithms may converge faster. However, we show that the resulting flows may be substantially different than the flows according to the max-min fairness. We further demonstrate that the max-min fairness solution can be very sensitive to small changes, i.e., there are configurations in which an addition or deletion of a session with rate δ may change the allocation of another session by Ω(δ · 2 n/2), but by no more than O(δ · 2n). This implies that it might be hard to locally estimate in a given state how close a session is to its max-min fair allocation.

Original languageEnglish
Title of host publicationProceedings of the 28th Annual ACM Symposium on Theory of Computing, STOC 1996
PublisherAssociation for Computing Machinery
Pages89-98
Number of pages10
ISBN (Electronic)0897917855
DOIs
StatePublished - 1 Jul 1996
Event28th Annual ACM Symposium on Theory of Computing, STOC 1996 - Philadelphia, United States
Duration: 22 May 199624 May 1996

Publication series

NameProceedings of the Annual ACM Symposium on Theory of Computing
VolumePart F129452
ISSN (Print)0737-8017

Conference

Conference28th Annual ACM Symposium on Theory of Computing, STOC 1996
Country/TerritoryUnited States
CityPhiladelphia
Period22/05/9624/05/96

Fingerprint

Dive into the research topics of 'Convergence complexity of optimistic rate based flow control algorithms'. Together they form a unique fingerprint.

Cite this