AT&Ts Salt Lake HFC Trial Seeks More Than Low Costs

AT&T Broadband & Internet Services has more up its
sleeve than merely another fiber-technology test with the trial of a new hybrid
fiber-coaxial design in Salt Lake City this fall.

If all goes according to the expectations of at least some
company strategists, the MSO could end up redefining how Internet protocol-based voice and
data services are transmitted over cable -- in effect outdating the existing cable-modem

But first, the company will have to demonstrate that the
basic HFC design concept -- dubbed "Lightwire" internally and widely known as
"mini-fiber node" -- is a practical option, AT&T Broadband spokesman Mark
Siegel said.

"This architecture is something we're testing in
anticipation that we're going to get millions of subscribers for our advanced
services," Siegel said. "If we do, the current architecture -- where 600
households are served from a fiber node -- may not be the best approach. But it remains to
be seen whether the new design is the right one."

The Salt Lake City trial, passing 66,000 homes and turning
on in October, follows a six-month study of multiple options for improving network
performance and capacity through segmentation of nodes. During that time, engineers from
AT&T Labs worked closely with the engineering team of the former Tele-Communications
Inc., led by chief technical officer Tony Werner.

Based on that study, the company believes it can extend
fiber deep enough to eliminate all in-line amplifiers on the coax network, thus reducing
serving-area sizes to under 100 households at a cost of $40 per home passed, AT&T Labs
district manager Xiaolin Lu said.

Describing the two-phase trial plans in a presentation at
the National Show in Chicago last month, Lu predicted that the operational savings from
lower power consumption and eliminating or reducing many maintenance procedures needed to
keep amplifiers properly tuned would add up to about $11 per home, per year.

These numbers alone, assuming that the cost projections are
right, could justify using the approach in systems that have yet to be upgraded to two-way
HFC status, AT&T Broadband district manager for product realization Marty Davidson

But Lu and Davidson noted that the real payoff on the new
design could be the means by which the freed-up bandwidth on the unamplified coax --
adding at least 250 megahertz to the 750 MHz now commonly available -- is put to use in
phase two of the trial.

Rather than delivering signals over that spectrum and in
the return in traditional modes, the MSO could use time-division multiplexing on an
end-to-end basis, allowing each serving area to operate like a local-area network.

"You can use the mini-fiber node to do some local
signaling, enabling very low-cost, very simple media-access protocols, versus having to
resolve everything back at the headend," Lu said. "You can utilize some kind of
Ethernet off-the-shelf product as the cable modem."

Such a capability would eliminate the requirements for the
complex media-access control, quality-of-service and other functionalities associated with
today's data delivery over HFC, thereby possibly extending the cost justification for the
deep fiber deployment to a much broader base of systems, including those already upgraded,
Davidson said.

"CPE [customer-premises equipment] is the biggest cost
factor in everything, so the payback there could be a major benefit," he noted.

Davidson said the design doesn't require the use of a
different fiber strand for every link to the mini-nodes from the fiber multiplexing node.

Options that could be used to allow a single fiber to serve
multiple mini-nodes include wavelength-division multiplexing and TDM, with WDM the likely
candidate. But Davidson was not certain what the choice was, and other officials could not
be reached for answers at press time.

There's also a strong likelihood that new types of
integrated optical transmitter/receivers now entering the market for use in
fiber-to-the-home applications in Japan and elsewhere could come into play in the new HFC
design, Davidson said.

This would greatly reduce the costs of the electronics at
each mini-node compared with what the optoelectronic costs at fiber nodes are today.

In order to accomplish the powering of the RF signal at the
mini-node across the full range of bandwidth, the system would use new
galium-arsenide-based chips, Lu said.

All of this adds up to what appears to be very high costs,
notwithstanding AT&T engineers' claims, said a skeptical senior industry engineering
executive who asked not to be named.

"Tony Werner says this looks doable, but we've seen a
lot of good ideas go up in smoke when it comes to putting them to the test in the
field," the executive added.

A case in point was the microcell concept used in
conjunction with delivering wireless personal-communications services over cable. "We
could just never make those numbers work," the executive said.

Nonetheless, if AT&T is right in its projections, the
TDM side of the equation would not seem to present any major technology hurdles, opening
the possibility of securing a robust means of delivering voice and data signals that would
overcome many of the technical challenges associated with today's cable-modem delivery

The question is whether the TDM approach would generate big
savings in terminal gear one year or two from now, when the volume demand for today's
cable-data technology will have had a chance to drive down the costs of modems and other

That's a question no one can answer at this point. But it's
clear that the cable-technology cost curve has a long way to go before it hits the numbers
associated with TDM-based technologies like 10BaseT Ethernet.