SAN FRANCISCO AT&T is developing a "broadband phone" that will be able to carry any type of packet traffic from voice to data to streaming video as specified by a touchscreen. Designed by AT&T's labs in Cambridge, U.K., the phone was described in a plenary session at IEEE Globecom on Tuesday (Nov. 28) by David Nagel, chief technology officer of AT&T and president of AT&T Labs.
The phone will fit into AT&T's packet strategy, which is based on a variant of a thin-client model that makes client devices as dumb as possible while keeping bandwidth assignment intelligence distributed in open packet networks. The phone uses a StrongARM 1100 processor to power a fully-programmable platform, comprised of a handset and a touch screen. The client assumes a constant 1-Mbit/second connection and uses a standard Ethernet interface model.
Early trials of the broadband phone are being conducted in tandem with a telecommuting application AT&T is testing, called a Remote Worker Application Server. This software suite places all functions of a telephone PBX on a home worker's PC, and allows Internet Protocol telephony to provide full advanced calling features and virtual private network applications. The original version of the software, designed around an analog telephone connectivity model, is being upgraded to handle coaxial-based voice-over-Internet Protocol services to align with AT&T Broadband's cable TV plans.
For reasons of both bandwidth efficiency and the pace of innovation, economics favor a rapid transition to all-packet traffic, Nagel told the plenary session.
Nagel said the model of what he calls "3G Internet" will blur the distinction between clients and servers in a network. The popularity of widely-distributed hosts, emerging in the Napster file-sharing model, indicates that "sooner or later, most clients will become servers, and this will lead to more symmetrical access patterns in the broadband Internet," Nagel said.
Path to profits
For this reason, advances in edge-access devices probably will be more important than advances in a photonically-switched core, Nagel predicted. It is not that edge alternatives like cable modems and digital subscriber line modems are not viable, Nagel said, but carriers have not figured out the best way to provision and maintain such networks to a large number of customers, and to make money in the process.
In this sense, moving to an all-packet infrastructure will carry some hidden bonuses, he said. If customers are billed on a flat rate or on an aggregate number of packets per time period, there will be fewer calls to customer support complaining of incorrect billing, Nagel said. It may take several years for the operational and maintenance advantages of an all-IP infrastructure to show up, he said, but the advantages are driving carriers away from old circuit-switched networks.