Stabilizing RED queue oscillation using the logistic map in AutoRED mechanism

UNCG Author/Contributor (non-UNCG co-authors, if there are any, appear on document)
Jiyoung Oh (Creator)
Institution
The University of North Carolina at Greensboro (UNCG )
Web Site: http://library.uncg.edu/
Advisor
Shan Suthaharan

Abstract: Active queue management (AQM) is one of the ways to control congestion at Internet Routers. One of the widely used AQM's is the random early detection (RED) scheme. The RED scheme suffers from chaotic queue oscillation problem particularly in a highly congested network. It causes jitter, high queuing delay when the queue size stays high, and underutilization when the queue size is low. Recently AutoRED algorithm has been proposed as a solution to the chaotic queue oscillation problem in that AutoRED calculates the weight, wq, continuously as opposed to a constant value set by a user [1]. AutoRED displays the reduction of the chaotic queue oscillation by network performance metrics and queue behavior graphs, but there has been no metric known to measure the degree of queue oscillation in terms of its effect on the Quality of Service (QoS). The purpose of the present study is twofold. Firstly, the possibility of an improvement by modifying AutoRED using a Logistic Map is investigated. This new technique introduces a user control parameter that can contribute to further improvements. Secondly, a new metric is proposed to show the degree of queue oscillation with regards to its effect on the QoS. The experiments are done by applying the new technique to network simulations in TCP only and TCP and UDP combined traffic environments. The results are compared with RED and AutoRED with regards to the proposed metric coupled with the network performance measurements and the statistical measurements of the queue behavior.

Additional Information

Publication
Thesis
Language: English
Date: 2009
Keywords
AQM, AutoRED, Logistic map, Oscillation metric, Queue Oscillation, RED
Subjects
Wide area networks (Computer networks)
Routing (Computer network management)
Queuing theory.

Email this document to