Evolutionary Optimal Topologies for Accommodating Traffic Growth

概要

論文の詳細を見る
Modern wide-area, Wavelength Division Multiplexing-based network topologies are often meant to have a much longer lifetime than it is possible to obtain using current technologies, as the greatest cost in the network deployment process is building the infrastructure itself. Thus, it is very important to design the topology in a way that supports traffic changes resulting from population migration, introduction of new services, but network failures as well. Networks obtained this way are more likely to be easier to extend using future technological advances. In this paper, we investigate a topology design problem for future WDM-based networks, using Genetic Algorithm with Lifetime Algorithm as an evaluation function, in search for the evolutionary optimal topology. Candidate networks are being developed in the process similar to the evolution of living organisms where every generation of new networks is being assessed for its ability to survive the longest, given certain, varying traffic load. Only positively-verified network topologies are allowed to pass to another generation, simultaneously being parents to descendants obtained from them in biologically inspired processes. We use Lifetime Algorithm for determining bottlenecks and detailed lifetimes of networks, together with the new application using it to develop new topologies for the sample geographical locations. We discuss the features of the networks that seem to be promoted and outline possibilities of further development of the method.
2011-07-14

Evolutionary Optimal Topologies for Accommodating Traffic Growth

スポンサーリンク

概要

著者

関連論文

スポンサーリンク