AT&Ts Success with LightWire May Lead to More Deployments

AT&T Corp.'s experimentation with a new fiber-rich
cable architecture appears to be panning out, leading officials to begin planning for
wide-scale deployments in 2000 and beyond.

"I think we'll be looking at this approach in a
number of locations," AT&T Broadband & Internet Services chief technology
officer Tony Werner said. "We have a ton of projects already under way, as well as
new ones where we think there's a role for this architecture."

The architecture in question, dubbed "LightWire
Neighborhood Broadband System," is in the pilot stage of first deployment over a
600-mile segment of AT&T's Salt Lake City network, about 35 percent of which is
now installed.

"We started lighting up the fiber last week, and we
will probably be cutting over some customers in the coming week," Werner said in a
recent interview.

He stressed that with the technical components of the
architecture already thoroughly tested, the primary purpose of the Salt Lake City project
was to determine whether his team's projected costs for upgrading systems to the new
architecture would pan out in the real world.

"At the front end of the project, based on completing
a couple of hundred miles of the construction, we're looking at cost parity with a
conventional 600-home node upgrade," he said.

"I'm not saying this architecture makes the most
sense for every situation," Werner added, "because when you have an inordinate
amount of underground plant, it may be hard to get this to prove in, at least in the near
term, owing to the amount of fiber you have to pull."

But it looks like LightWire will prove in for upgrades
involving plant where the ratio of aerial-to-underground plant is "the typical 70-30
split," he said.

The Salt Lake City system offers a glimpse at how widely
applicable the new architecture might be, insofar as this is not an antiquated,
low-capacity system, but rather one that was state-of-the-art as of the beginning of this
decade. The system's capacity is 550 megahertz, activated one-way, with some fiber
already in use, Werner noted.

LightWire is designed to extend the core techniques used in
AT&T's current, state-of-the-art, 600-home-node, 750-MHz systems to an
architecture that pushes fiber to the point where all in-line amplifiers beyond the node
are eliminated, resulting in an 80 percent reduction in field electronics.

In place of the typical fiber node, the architecture uses a
"mux node," which serves as the interface between all of the distribution fibers
and the nearest hub. The company is extending two fibers -- one for each transmission
direction -- from the mux hub to the mini-fiber nodes that connect with the coaxial plant.

"We might go to one fiber using course WDM
[wave-division multiplexing]," Werner said, in reference to the two-wavelength
combination over fiber made possible by operating in the 1310-nanometer and 1550-nm

Despite the apparent jump in the amount of fiber required
to go from 600-node configurations to ones where the average home count is 80 per node,
the actual ratio of fiber-to-coax plant miles in the LightWire design is only 30 percent,
versus 15 percent to 20 percent for the typical 600-home node design, Werner said.

Moreover, he added, the cost of fiber has been falling at a
rate of 20 percent per year, and costs for optoelectronic components are dropping even

While there may be no immediate net savings by upgrading to
LightWire specs, the fact that it can be done without breaking current budget allotments
gives AT&T a long-term benefit both in terms of cost savings and the ability to
migrate to more advanced services and higher levels of market penetration for such

Most important in the near term, the passive design
significantly reduces ongoing maintenance costs and cuts power consumption, Werner noted.

As for how the architecture addresses future needs, Werner
asserted that at this time, the company does not feel pressed to go beyond the capacity
potential of the 600-home-node topology, which is why it can begin implementing the
LightWire design in projects that have already upgraded substantial portions of a system
to the 600-home node architecture.

"We have so much capacity in the 600-home design that
we will operate the LightWire segments as if they were 600-home areas, as well, for a long
time to come," he added.

But there is a second phase to AT&T's LightWire
strategy, which is still being refined for eventual testing at some point starting next
year, Werner said.

In that phase, the company will look at how it might take
advantage of the added coaxial bandwidth that comes with eliminating amplifiers --
including ways to use the very high end of the spectrum for transmitting new packet-based
services in both directions as an overlay to the existing service structure.

For now, though, the emphasis is on doing things that fit
into the current state of the art with respect to how fiber nodes operate and data
services are distributed.

For example, the mini-fiber nodes being supplied by Corp. for the Salt Lake City project are miniaturized refinements of the
fiber-node technology the company has been supplying to the industry for traditional
hybrid fiber-coaxial systems, CEO David Woodle said.

"The way this works is pretty common to today's
architecture," Woodle noted. "The fiber comes into an optical splitter with
multiple outputs that feed line extenders to send the signals out in different directions
over the coax."

The forward and return paths work the same way. But because
the four coaxial transmission paths are so close to the end-users, no amplification is
required beyond the mini-fiber node. "We don't see this as a major deviation
from what C-COR has done for a number of years," Woodle said.

He added that it was possible that greater economies could
be achieved with the LightWire architecture if AT&T could use one fiber transmitting
several wavelengths to serve several mini-nodes from the mux node.

Research into the wave guides and other components
associated with the use of dense WDM at this end of the cable system is under way at and at Silicon Valley Communications Inc., a Santa Clara, Calif.-based supplier
of fiber optic transmission systems that is acquiring.

Other firms involved in the supply of systems designed
specifically for the LightWire project include Harmonic Inc. -- which, along with SVCI, is
providing mux nodes -- and General Instrument Corp., which is supplying mini-fiber nodes,
Werner said.

In addition, Scientific-Atlanta Inc. is working on new
optoelectronic components that could be used in various applications of DWDM and other
approaches to optical multiplexing.

S-A recently invested in U.K. optoelectronic-components
firm Bookham Technology Ltd. to help speed adaptation of that firm's new
integrated-device technology for applications in cable.