A hybrid genetic algorithm/variable neighborhood search approach to maximizing residual bandwidth of links for route planning

Gajaruban Kandavanam, Dmitri Botvich, Sasitharan Balasubramaniam, Brendan Jennings

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

8 Citations (Scopus)

Abstract

This paper proposes a novel approach to performing residual bandwidth optimization with QoS guarantees in multi-class networks. The approach combines the use of a new highly scalable hybrid GA-VNS algorithm (Genetic Algorithm with Variable Neighborhood Search) with the efficient and accurate estimation of QoS requirements using empirical effective bandwidth estimations. Given a QoS-aware demand matrix, experimental results indicate that the GA-VNS algorithm shows significantly higher success rate in terms of converging to optimum/near optimum solution in comparison to pure GA and another combination of GA and local search heuristic, and also exhibits better scalability and performance. Additional results also show that the proposed solution performs significantly better than OSPF in optimizing residual bandwidth in a medium to large sized network.

Original languageEnglish
Title of host publicationArtificial Evolution - 9th International Conference Evolution Artificielle, EA 2009, Revised Selected Papers
Pages49-60
Number of pages12
DOIs
Publication statusPublished - 2010
Event9th International Conference on Artificial Evolution, EA 2009 - Strasbourg, France
Duration: 26 Oct 200928 Oct 2009

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume5975 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference9th International Conference on Artificial Evolution, EA 2009
Country/TerritoryFrance
CityStrasbourg
Period26/10/200928/10/2009

Fingerprint

Dive into the research topics of 'A hybrid genetic algorithm/variable neighborhood search approach to maximizing residual bandwidth of links for route planning'. Together they form a unique fingerprint.

Cite this