NeTS: Small: Migration Towards Terascale Flexible-Grid Optical Networks

University Of California-Davis, Davis CA

Investigators

Abstract

Fiber optic networks are deployed worldwide in telecom backbone networks to carry the huge and exponentially growing amount of Internet traffic. However, this traffic growth is straining the capacity of fiber optic networks, which has led researchers to explore new ways of using the bandwidth available in them. Traditionally, a fiber's spectrum is divided into a fixed grid of equally-spaced optical channels. To squeeze more out of the available spectrum, this conventional fixed grid is evolving towards a flexible grid, where individual channels can be placed anywhere in the spectrum, and sized to match the traffic they carry. This research will investigate novel migration strategies from fixed-grid towards next-generation flexible-grid fiber networks, and strategies for maximizing the traffic a network can carry. This will enable high network efficiency, thus lowering the cost of the communications networks that underlie the Internet. Flex-grid technology holds enormous potential, but also brings with it significant complexity. This project tackles the two main issues associated with incorporating flex-grid technology into communications networks. First, it will investigate potential fixed-to-flex grid migration strategies, taking into account the complexities of device characteristics, detailed node architectures and physical (signal) impairments in network optimization studies. The project will systematically address possible upgrade paths, including which node(s) should be upgraded first, whether and where flex-grid 'islands' should be created within networks, and how many nodes should be upgraded for current network state. The project will use a 24-node nationwide network with aggregate terabit-per-second traffic demand to explore the relative advantages and disadvantages of each approach. The project's second major thrust is to tackle the problem of traffic grooming in such networks. While traditional traffic grooming is essentially a combined routing and (fixed-size) bin-packing problem, the focus here is on novel strategies to efficiently aggregate traffic as container (bin) sizes vary in time.