CNET también está disponible en español.

Ir a español

Don't show this again

Christmas Gift Guide
Mobile

AT&T revises cable telephony tests

The long distance giant revises its early rollout plans for its cable telephone system, adding a new set of tests aimed at reducing network traffic jams.

AT&T is revising the early rollout plans for its ambitious cable telephone system, adding a new set of tests aimed at reducing network traffic jams once the technology becomes widely used.

The company is in the first stages See related story: The new world order of testing technology for cable telephony, with the hope that the trials will help it directly compete with local phone companies like Bell Atlantic and US West. Ultimately, AT&T wants to offer consumers a one-stop shop for local and long distance voice service, high-speed Internet, and cable TV.

The public phase of its cable phone tests began earlier this month in Fremont, California--the first of ten cities scheduled to try out the service this year.

But the company is now throwing a new element into its tests, using a radically different network configuration in a Salt Lake City trial that could help the company maintain call quality as more customers sign up for the service.

AT&T says that the Fremont tests have been going well, and that the addition of Salt Lake City is an unrelated test of new technology. Test for other cities are proceeding along an original plan, unaffected by plans for Salt Lake City, a spokesman said.

"The technology we're using in Fremont and the other cities works very well. We're very pleased," said Mark Siegal, an AT&T spokesman. "We are not redesigning our cable telephony network, nor are we changing the costs associated with it."

But the new trial, which would vastly scale back the number of users sharing each piece of the cable system, could help the company preserve the quality of its telephone service, analysts said.

Testing the system's limits
The technology that AT&T is using in Fremont and other cities has been well tested in the field already, analysts say.

Cox Communications, the current U.S. leader in cable telephony, already has more than 42,000 users on its system and uses similar architecture to AT&T's, said Michael Harris, president of cable research firm Kinetic Strategies .

Under those systems, an average of between 500 and 1,000 homes would share a single cable network. AT&T says each individual strand of its network will pass about 600 homes.

This isn't a problem, however, until close to 25 percent or 30 percent of these homes passed sign up for the same services, and use the system at the same time, analysts say. Once this happens, the quality of calls could degrade, potentially creating a echo-like interference similar to that experienced on some cellular calls.

The problem can be compared to what happens with broadband cable Internet services such as @Home. Users share the amount of bandwidth available on a neighborhood cable network--so a larger number of customers online decreases the bandwidth available to any single user.

The call-quality issue is not universal, however, and can have more to do with the quality of a given cable network than the number of users on a system. "We have nodes well in excess of 30-percent penetration, and we're not experiencing that," said Chuck McElroy, Cox's vice president of new services support.

AT&T's cable infrastructure is generally viewed as less advanced than companies like Cox or MediaOne, which have already moved heavily into cable telephony, analysts say.

Cable companies can fix their bandwidth problem fairly simply. By adding equipment that splits a cable "node" serving 600 or so customers, an operator can divide a system so that there is the equivalent of 300 users on two seemingly separate systems, for example.

AT&T is essentially testing an advanced version of this technology in Salt Lake City, adding fiber-optic lines deeper into the cable system and creating networks shared by just 50 to 75 people.

This kind of configuration would be more expensive to roll out than AT&T's planned telephony network. But even if it is successful, the company wouldn't need to implement this kind of system for years, analysts said. If bandwidth begins to be a problem, the company could use more incremental "node-splitting" upgrades, or simply dedicate more of their cable channels to telephony, Harris said.

Nevertheless, the Salt Lake City trial could be a kind of second-generation test, aimed at gathering data for use if the system does become popular faster than expected, said Yankee Group senior analyst Brian Adamik.

"This is a brand-new network they're rolling out, and there will be bumps along the road," Adamik said. "That's why you have trials. There are new market findings every day."