News Analysis: AT&T's Embrace of New Technology
Signals Next Era
By SETH SCHIESEL
mid last week's typically hectic flow of headlines about the communications industry, perhaps
the most significant news of all got lost: Frank Ianna, president of the AT&T Corp.'s network
unit, announced that by the end of this year, the long-distance giant wanted to stop buying traditional
telephone switches for the core of its network.
Granted, Ianna's pronouncement last Tuesday was not the lapel-grabbing kind of news generated by
a multibillion-dollar takeover or the latest hot Internet stock. But it was a seminal moment for the
communications industry and in some ways for everyone who uses a phone.
As it halts decades of investing in the traditional gear that makes up most of the global phone system,
AT&T is retooling its network around a new generation of technologies inspired partly by the
Internet.
Sprint and MCI Worldcom, the other long-distance giants, are doing much the same thing. But as
the biggest communications company in the United States, AT&T wields a proportionately larger
influence in the market and with technology suppliers.
So Tuesday's announcement was perhaps the definitive sign that at least in the long-distance
business, the aging and increasingly crowded phone network would be rejuvenated with younger,
more efficient technologies.
For consumers, this evolution could lead to lower prices. And in the future, as the new technologies
expand from the core of the network to homes and businesses, consumers could also benefit from
new sorts of features and services.
"As the functionality moves closer and closer to the customer, the bottom line for the consumer is
that clearly we will drive the economics down," said Neil J. Grenfell, an engineering vice president at
Sprint. "This can also help give the customer more and more control and more and more integration."
This might mean, for instance, that a tourist or business traveler who wanted to stay in touch could
simply unplug the handset from the home or office telephone and plug it into the base of any other
telephone. The network would recognize that the person had moved around the block or across the
country, and calls to the original number would ring in the new location.
Services like that are many years away and would require the local phone companies to invest
billions of dollars on top of the current spending by long-distance carriers. And, of course, it would
require consumers to buy "intelligent" phones.
The reason the new advanced networks being planned by AT&T and others are new and advanced
is that they change the way networks behave at their most basic level.
Since the invention of the telephone, almost every civilian communications network has been based
on the concept of circuits. If a network is a multilane highway, then a system based on circuits,
known as a circuit-switched network, paints the lane lines of that highway solid: each conversation
has its own lane, or circuit. If two people are on the phone with each other and neither of them is
speaking, they still use the entire circuit just as if they are screaming. The standard voice circuit sends
and receives 64,000 bits of information a second.
The main advantages of circuit switching are that it is reliable and relatively simple to carry out. The
main disadvantage is that it is extremely inefficient to keep an entire lane open for every phone call,
regardless of how much data -- in this case, conversation -- it is carrying.
The alternative that is being embraced by AT&T and the others is a sort of mass-transit alternative
called packet switching. On a packet highway, the lane lines are dashed. Rather than consuming at
entire lane, each conversation gets broken up into millions of small pieces, or packets, that are then
mixed up with other conversations only to be reassembled into intelligible communications at the
other end.
The main advantage of packet switching, the technology used by the Internet to move massive
amounts of data around the globe, is that it is extremely efficient. A conversation that takes up
64,000 bits a second on a circuit-switched network might take up only a quarter of that, or 16,000
bits a second, on a packet-switched system. That can allow carriers to lower their costs and,
potentially, the prices they charge consumers.
The problem is that sorting out all those packets and routing them to the right place requires
ingenuity, especially when the packets are carrying conversations. A few extra seconds when loading
a Web page may not mean much, but consumers expect their phone calls to be perfect
representations of what they are saying -- with no delays or distortion.
The means for making packet-based systems robust enough to carry millions of phone calls has just
emerged over the last year or so from companies including Ascend Communications, which has
agreed to be acquired by Lucent Technologies, Cisco Systems and others.
Packet-based, however, does not necessarily mean the Internet version, known as Internet
protocol.The main technology that AT&T and many other long-distance carriers are looking toward
is called asynchronous transfer mode.
In the jargon of the communications industry, ATM uses "cells" rather than packets. That is because
while the data packets generated by Internet protocol vary in length, the data cells in an ATM
system are always the same size, generally 53 bytes. That uniformity can be a powerful advantage
because it allows ATM to carry co-mingled voice, video and data traffic without sacrificing much
efficiency.
Also, while Internet protocol often ships each packet of information to its destination by a different
path, ATM sets up so-called virtual circuits, which help insure that a transmission of Hamlet's
soliloquy begins with "To be" instead of "Be to" or "Eb ot."
Sprint is basing its network of the future largely on ATM switches made by NEC of Japan, though
Sprint is also working closely with Cisco. AT&T is using Ascend's ATM equipment. MCI
Worldcom has asked vendors to submit proposals for ATM switches but has not yet made a
decision.
One of the raging debates in the communications industry is whether ATM is necessary at all or
whether Internet protocol is reliable enough to form the underlying network language instead. For
now, most of the big carriers are wary of Internet protocol, partly because not all of the work has
yet been done to add advanced features like Caller ID and call-waiting to Internet-based systems.
"Think about the software you need to do call-waiting in the circuit world," Ianna said. "Think about
the software and the intelligence you need to replicate that in the IP world. It's non-trivial. There is a
lot of router work that needs to be done, and it's not there yet."
The last time AT&T revamped its network so thoroughly was about 10 years ago when it upgraded
its long-distance network from copper wires to optical fiber. About 20 years ago, the company
upgraded its circuit switches from analog to digital.
''This is the next-generation architecture," Ianna said. "If analog was the first generation and digital
switching was the second and digital transmission was the third, this is the fourth. It is significant."
|
