NeTS: JUNO: Cost-Effective and Scalable Architectures for Multi-Granular Optical Networks

George Washington University, Washington DC

Investigators

Abstract

Optical networks have expanded their reach from national backbone networks to metropolitan and customer access networks, and are making their way into datacenters. This has been made possible due to great strides in two arenas: optical switching technology including the development of reconfigurable optical add-drop multiplexers and wavelength-selective switches; and efficient transmission methods that enable flexible provisioning of elastic bandwidth to applications. Nevertheless, there are fundamental fiber capacity limitations that will inevitably necessitate multiple parallel fibers between switching nodes and larger switches. A solution to the switch-scaling problem is to bundle some of the wavelengths or frequency slots together and switch them as a band. Yet, current methods are inadequate because technological advancements have rendered some of the research obsolete or have introduced new problems that have not been attempted before. The objectives of this joint US-Japan project are to develop novel scalable and cost-effective multi-granular switching node architectures, and to develop efficient algorithms for provisioning bandwidth to applications. Scaling an architecture typically introduces some form of penalty, such as a provisioning constraint, that the network operator must respect. This project will aim at identifying what features bring about the most cost reduction and increase in scalability without overly limiting network operators? flexibility. The architectures and algorithms will be evaluated using theoretical analyses, simulations, and small-scale experiments. By targeting an important component of the cost of optical networks, this project will enable the faster adoption of optical infrastructure and the availability of high-speed networking for all.