Abstract

The recent advent of commercially deployed optical networks employing all-optical switching and ultra-long-haul transmission has re-ignited interest in the debate over the role of wavelength conversion.^1–3 Such networks take advantage of optical bypass, where a signal can pass through a network node in the optical domain without being electronically terminated, to eliminate much of the electronics required in the network nodes. This significantly reduces the cost of the overall network, as well as the space and power requirements, and greatly simplifies and accelerates the provisioning process.^4,5 Optical bypass does impose a wavelength continuity constraint: if a signal is transmitted all-optically over a number of hops, it must use the same wavelength on each of the hops (assuming all-optical wavelength conversion is not available). While it is possible to come up with scenarios that show this constraint on wavelength assignment results in a loss of network efficiency, it is important to consider the practical architectures in which ultra-long-haul all-optical systems will be deployed. High capacity, all-optical ultra-long-haul systems are best suited for express backbone networks, which we refer to as all-optical express networks.

© 2002 Optical Society of America