HKUST Library Institutional Repository Banner

HKUST Institutional Repository >
Computer Science and Engineering >
CSE Master Theses  >

Please use this identifier to cite or link to this item: http://hdl.handle.net/1783.1/5758
Title: Active queue management schemes using a capture-recapture model
Authors: Chan, Ming Kit
Issue Date: 2002
Abstract: To provide fair bandwidth sharing for TCP flows, the Internet depends on the cooperation between TCP sender and subnet routers to adjust source data rates in the presence of network congestion. On the other hand, Active Queue Management (AQM) schemes are used in the Internet routers to protect TCP flows against non-responsive flows. As a result, one of the major objectives for an Internet router is to allocate these flows in order to achieve maximum fairness. However, most of the existing AQM schemes cannot provide accurate fair bandwidth share. In this thesis, we propose a new AQM scheme called CAP (CAPture-recapture fair sharing) to solve this problem. The CAP algorithm estimates the source data rates and the fair share by the random captures of the incoming packets. The CAP algorithm is based on a capture-recapture (CR) model, which was originally developed for estimating the size of animal populations. Conceptually, the CR model depends on several key ideas: animals are captured randomly, marked, released and then recaptured randomly from the population. We have applied the same approach to the AQM schemes by capturing packets in the buffer. The scheme is simple to implement, robust, efficient, and does not require keeping per-flow information. Using analysis and simulation, CAP is shown to provide highly accurate fair bandwidth share under different network configurations. Our experiments also show the CAP algorithm outperforming the existing mechanisms. Finally, we implement our algorithm in a software router. The results produced by using real network traffic conform to the simulation results.
Description: Thesis (M.Phil.)--Hong Kong University of Science and Technology, 2002
x, 62 leaves : ill. ; 30 cm
HKUST Call Number: Thesis COMP 2002 Chan
URI: http://hdl.handle.net/1783.1/5758
Appears in Collections:CSE Master Theses

Files in This Item:

File Description SizeFormat
th_redirect.html0KbHTMLView/Open

All items in this Repository are protected by copyright, with all rights reserved.