The prospects for optical networking are about to change
dramatically due to the strength of new trafficking components supporting multiwavelength
architectures in telecommunications-distribution backbones.
Where today, the use of multiple wavelengths over single
fibers has become a mainstay in long-haul networks, and it is beginning to penetrate
distribution backbones as a means of enhancing traffic-bearing capacity in point-to-point
links, the next breakthrough -- where traffic is routed, monitored and protected in the
optical domain -- promises to free carriers from the electronic bottlenecks that hold back
full realization of the fiber bandwidth potential.
"The important thing to remember is that if you want a
true, all-optical backbone, you need the functionality that allows you to provision
wavelengths dynamically, as well as to perform the monitoring and other tasks associated
with traditional network architectures," said Farooque Mesiya, CEO of Tellium Inc.
Tellium -- a start-up backed by SAIC Inc., the new owner of
Bell Communications Research, as well as by laser-supplier Ortel Corp. and several
investment firms -- is in the vanguard of this new generation of optical networking, with
plans to bring its first products to market during the second quarter, relying in part on
patented technology developed at Bellcore.
The firm, based in Edison, N.J., recently won a contract
from the Department of Defense's Advanced Research Projects Agency (DARPA), in
competition with major suppliers, for the second ring of an all-optical network connecting
government agencies throughout the Washington, D.C., area.
"Basically, this is a network with a predefined number
of wavelengths [16] that will be using our transport and cross-connect elements,"
Mesiya said. He also noted that the equipment that Tellium is supplying is designed to
support expansion to 32 wavelengths.
The firm's 32-wavelength transport system will be
available in the second quarter, and the first iteration of its cross-connect will come
online by midsummer, with capacity expanding to 128 bidirectional ports by year's
end. There will be other elements in the firm's product line, as well, he added.
While a handful of vendors introduced optical add/drop
multiplexers during the past year, these devices are not true cross-connects, said Tim
Krauss, director of product marketing and business development for Tellium's
lightwave-products unit.
"To be a real cross-connect, it must be nonblocking,
which none of the so-called optical cross-connects being introduced now can do," he
said, noting that Tellium intends to introduce true optical cross-connect next year.
With the nonblocking capability, not only are the
wavelengths routed or switched on the fly, but they are translated into other wavelengths
to take advantage of available wavelength slots in a given fiber link as those slots open
up in the ongoing routing process. Without this capability -- which must be accomplished
without converting the signal to electronic frequencies and then regenerating it -- a
switched wavelength will be blocked until its slot opens over a desired link, which is
something that carriers can't live with in the high-volume trafficking environment of
local telecommunications.
Alcatel is supplying cross-connects supporting four to
eight wavelengths as part of the multicarrier, multivendor Advanced Communications
Technologies and Services project in Europe, Krauss said. Commercial products will support
more wavelengths over more ports, probably in 128-by-128-port configuration, he said.
Already, he added, the prospects associated with
next-generation optical gear are shaking up carrier assumptions about the functionality
and business uses of their networks.
For example, he said, owners of fiber links will be able to
sell "fractional dark fiber," using some wavelengths for themselves and
wholesaling others. Moreover, the marketing of transport on a wavelength basis will allow
end-users and operators to implement services as needed without investing in equipment
that will aggregate the signals carrying those services into higher-speed backbone tiers,
such as high-speed SONET (synchronous optical network) layers.
"You can add a fast Ethernet link without having to
choose between mapping it into something else or using separate facilities," Krauss
said, noting that with deregulation and unbundling of telco plant, telcos have an
additional incentive to create multiple paths using DWDM (dense-wave-division
multiplexing) over existing fiber plant.
Alcatel has been working with electro-optical
semiconductors as one promising avenue of development for all-optical routing systems, but
there's no clear winner yet among the technical options that it and other
manufacturers have for building the cross-connects.
For example, Lucent Technologies -- which brought an
add/drop multiplexer to market last year, and which is the supplier of the first ring in
the Washington DARPA project -- is looking at optical waveguides that alter wavelength
characteristics in response to electrical impulses and at micro-electro mechanical systems
devices, which use tiny, movable mirrors to route traffic.
Researchers at Lucent's Bell Labs unit have developed
a device that uses a diffraction grating to separate multiple wavelengths coming in from a
single fiber and then passes each wavelength to a separate micromirror that is
mechanically turned by an electrical impulse to determine the routing direction of each
wavelength.
The group, which has demonstrated the device in a
four-by-four-port array with 16 wavelengths, is working on a system that can handle a much
greater volume of wavelengths at potentially lower costs than would be the case using
other techniques for high-density wavelength trafficking, said Joe Ford, a researcher on
the project.
"We're not at the product-development stage
yet," Ford said, noting that MEMS devices are already widely used in automobile
airbags. "There are other technologies competing with this one, but the advantage
here is that you can scale cost-effectively to high densities of wavelengths."
Mesiya declined to go into technical details regarding
Tellium's approach to cross-connect design. But he made it clear that the issues of
scalability and dynamic wavelength provisioning cited by Krauss and Ford were priorities
in the firm's product development.
"This technology is now at the stage where we can
truly make a fiber backbone network," Mesiya said. This entails remote, dynamic
provisioning of wavelengths, monitoring of the bit-error rate and other performance
elements and restoring of the traffic at a high level of grooming in the central office,
without requiring that the signal be broken down and reconstituted, he noted.
"Our view is that in the all-optical domain, SONET
migrates to the edge of the network, because you have dynamic control of multiplexed
signals at the core," Mesiya said. "If you have a data router that is putting
out a multiplexed signal at OC-48 [2.48 gigabits per second], you don't need to run
it through a SONET multiplexer if you're operating over a purely optical
network."
Much as the cable industry has long sought ways to
distribute video signals digitally without adding the overhead of SONET, so will the
increasingly data-packed networks of telecommunications firms of all types, including
cable, want to avoid unnecessary overhead, Mesiya noted. Such capabilities not only mean
that networks will realize major economies from the capacity expansion provided by DWDM,
but that they'll also save money by eliminating many of the electronic elements that
are essential to operating in today's environment.
