Making a higher jump to light speed, As Sonet OC-192 rolls out, vendors work on OC-768 >BY BETH SNYDER, Associate Editor-News

It's just like the saying, "You can never have too many friends or too much money." Except in the telecommunications world, you can never have too much bandwidth.

Just a few years ago, people thought Sonet time division multiplexed (TDM) OC-48 at speeds of 2.5 Gb/s was incredible. Then came wave division multiplexing (WDM), which multiplied those 2.5 gigabits by two, four, eight and even 16. Earlier this year, OC-192 at 10 Gb/s TDM was introduced commercially and is starting to show up in networks.

"It's unbelievable how fast bandwidth demands are growing," said Philippe Moran, Northern Telecom's senior manager for optical networks. "We never thought we'd introduce two channels on OC-192 this year, but as soon as we brought out 192, customers asked for two channels.

It's inevitable that the discussion turns to the next step in TDM: OC-768 at 40 Gb/s.

As it turns out, many vendors are already hard at work in the lab and discovering early problems. For example, chromatic dispersion on the fiber at a super high speed dramatically drives down the span distance. Also, dispersion compensation management on 40 Gb/s TDM has to be much more exact than it is now to handle the souped-up speed.

New fiber could solve the problems, but it is expensive for carriers to rip out old fiber and replace it. It is much more cost-effective to use what is already in the ground.

Another problem is that the available electronic optical modules can't create 40 Gb/s pulses to send down the fiber. That, of course, will probably be solved as electronic chips get faster and continue to evolve.

"Anybody who has ever predicted that there isn't going to be a demand for bandwidth has always been wrong," said Tom McDermott, vice president and chief technical officer at Alcatel Telecom. "I see WDM and multiple carriers as good starts for now.

So once again, the industry comes back to the recurring question of TDM vs. WDM. Although most large Sonet vendors lean toward a TDM solution, most of them also see WDM playing a role in future speed ramp-ups.

"It's just like the battle between OC-48 and OC-192. My view is that eight channels of OC-192 is going to be cheaper than two channels of OC-768," Moran said.

Before that changes, two things will have to happen, he said. The potential volume of 40 Gb/s will have to increase, and new inventions in fiber optics and electronics will have to be discovered.

Fujitsu announced a successful trial of 40 Gb/s speeds using 20 Gb/s TDM with two-channel WDM earlier this year.

"Fujitsu believes in most cases that, for efficiency and network management, it's better to go to straight 10 Gb/s TDM system vs. four-channel 2.5 Gb/s," said Hal Calhoun, senior manager of broadband transport planning at Fujitsu. "Going forward, it's hard to say. We're definitely starting to see a bigger hybrid of TDM and WDM. One thing is for sure: 10 Gb/s is not what they ultimately want-10 Gb/s is just what they'll take for now.

Kathy Szelag, Lucent Technologies' director of Sonet product planning, said Lucent is developing a 40 Gb/s system in the lab with prototype boards and lasers. But because WDM has caught on quickly and is relatively inexpensive, there simply isn't a future where either TDM or WDM wins.

"Two years ago, no one knew how to build an affordable WDM system," she said. "Now they're about equally cost-effective. And the reality is that in three, four or five years, we'll see about the same amount of WDM sold as TDM. I don't see one slowing down the other.

As far as OC-768, no one can predict when it may become reality. Even if all the problems were fixed and manufacturers started to build today, it would be a two-year project, Szelag said.

She said it's more likely about four years away. Other vendors agreed-about four years has been the traditional cycle of TDM speed upgrades in the past with OC-192 and OC-48.

Of course, not everyone agrees that 40 Gb/s TDM is a sure thing.

Mark Lutkowitz, president of Trans-Formation, a Birmingham, Ala.-based consultancy, said the problems of OC-192 are already great enough. WDM at 16 and even 32 channels is a more likely solution for higher speeds, he said.

"There are suppliers saying OC-768 will be available in the year 2000. I just don't believe that," he said. "People will go to OC-192, but then what after that? I think it'll be a very long time, if ever, for 40 Gb/s TDM.

Either way, bandwidth demand will continue to march on and compel researchers to look at all the possibilities. As McDermott said, "It's definitely a hot topic of research. Work is going on in higher data rates in labs all over the place.