LightWire Pushes Into Phase 3

AT&T Broadband & Internet Services has pushed its
thinking about the evolution of its passive "LightWire" hybrid fiber-coaxial
architecture into a third phase, in which highly integrated optoelectronic components and
digital technology will radically alter the way services are delivered today.

The three-phase approach to implementing LightWire was
outlined by AT&T Broadband vice president of engineering Oleh Sniezko at the recent
2000 Conference on Emerging Technologies sponsored by the Society for Cable
Telecommunications Engineers.

While elements of the strategy have been known for some
time, Sniezko's description provided new insight into the company's thinking about using
digital-baseband and dense-wavelength-division-multiplexing technologies and about the
evolution to what it calls a "distributed CMTS (cable-modem-termination-system)"

"We want to do distributed processing," Sniezko
said. "I'd guess that in five years, distributed CMTS will be used in 80 percent of
our systems."

The idea is to trim functions at the primary hubs to
bundling and routing, while moving MAC (media-access-control) functions to the mini-nodes
in the LightWire architecture. These points of final fiber termination serve about 70
homes passed via passive coaxial cable -- coax that has no in-line amplifiers between the
node and the end-user.

Each mini-node will handle local policing and resolve
upstream contention within the serving area, independent of the rest of the network,
Sniezko explained. This will require converting signals to the time-division-multiplexed
baseband domain, first in the upstream and eventually in the downstream.

Already, Sniezko noted, Broadcom Corp. and other chip
suppliers offer 100-megabit-per-second and gigabit Ethernet chips that would make it
possible to interact with end-users on the coax as if they were service nodes on a
local-area network.

"We want to push the HFC [hybrid fiber-coaxial]
network to look exactly like a PON [passive optical network], while still using the
coax," Sniezko said.

PONs -- which, in a telecommunications environment,
typically use fiber-to-the-curb or fiber-to-the-home architectures -- use several
techniques to transport signals, including TDM, DWDM and subcarrier multiplexing,
depending on carrier and vendor designs. "We want to use all of these technologies to
our advantage," he added.

The game now is to plot the migration strategy, which has
evolved from a two- to a three-phased approach in the company's current thinking.

As AT&T moves into the first phase of LightWire
upgrades -- retaining current modes of signal delivery in the downstream and upstream
paths -- the fast-falling cost curves are opening opportunities to make more efficient use
of this architecture through changes in those delivery modes.

The first such shift can already be seen in adopting
baseband-return signaling between secondary and primary hubs, which AT&T expects to
extend to nodes and mini-nodes over the next year or so.

In phase two -- the timing of which is still to be
determined -- using digital baseband for dedicated signals in the downstream, as well as
upstream, will allow the MSO to "daisy-chain" mini-nodes via a two-fiber strand.

One fiber carries the broadcast analog signals, and the
other operates as an OC-48 (2.5-gigabit-per-second) bus, allowing use of TDM devices to
add and drop signals from the bus at each mini-node.

This daisy-chaining makes adding mini-nodes easy and
greatly simplifies management and restoration of services, Sniezko said.

"LightWire III" would entail a shift to the use
of DWDM to deliver dedicated signals to each mini-node in the daisy chain.

This version of LightWire would rely on very low-cost LEDs
(light-emitting diodes) and a new generation of integrated optoelectronic devices to
create an all-passive environment from the secondary hub all the way to the mini-nodes,
Sniezko said.

"If the network capacity is enough between the primary
and secondary hubs, the narrowcast part of the [traditional cable] subcarrier-multiplexing
system can disappear," he noted.

The architecture would use DWDM to deliver dedicated
baseband signals to secondary hubs, then use DWDM at secondary hubs to partition signals
across multiple wavelengths, so each wavelength would carry signals dedicated to a
specific mini-hub.

One could use inexpensive LEDs at the secondary hubs,
rather than high-cost wavelength-specific lasers, by employing optical add/drop
multiplexers to deliver a "slice" of each LED's output over a narrow wavelength
into the fiber, Sniezko explained.

The key is a new generation of optical devices that can be
tightly integrated onto electronic circuits. "The enabling devices exist today, but
that must be packaged to our specifications," he said.

Such packaging is under way through the work
Scientific-Atlanta Inc. is doing with Bookham Technology Ltd., a U.K.-based developer of
such devices.

"We're still a long way from the ultimate in
solid-state optical networking, but we're moving a good way up the scale from where we've
been," Bookham vice president of business development Robert Green said.

Bookham developed a way to form complex optical circuits on
mass-produced silicon chips, with the potential to miniaturize and cut costs of products
to be used in DWDM, return-path transmissions and fiber optic nodes.

An early Bookham product is an integrated
transmitter/receiver being used by Japan's Nippon Telephone & Telegraph Corp. and
other entities around the world to help lower the cost of FTTH systems.

Bookham -- backed by the likes of Cisco Systems Inc. and
Intel Corp. -- is working with S-A to develop a number of cable applications for its
patented "ASOC" (application-specific optical-circuit) technology, Green said.

One application the two companies are exploring is
integrating a wavelength demultiplexer and photoreceiver onto the chip. This would create
a miniaturized, low-cost means of handing off a wavelength from a multiwavelength stream
at a mini-node on the cable plant -- which is what AT&T wants in the third phase of

Early results of the Bookham/S-A collaboration suggested
that some existing capabilities in Bookham products can be readily transferred to cable
applications, possibly this year, Green said. But the DWDM add/drop capability is not one
of them.

AT&T officials acknowledged that they are closely
following the S-A/Bookham developments.