Optical components

Kansans Reset the Clock

Remember the days when all-optical networks were being touted as a way for carriers to slash costs and make their infrastructures future-proof?

One of the big stumbling blocks was finding a way to perform “3R” regeneration of optical signals -- reamplification, reshaping, and retiming (see Optical Detection & Regeneration ). The biggest challenge was retiming, restoring the “clock” used to identify the positions of the ones and zeroes at the end of a long transmission line.

A bunch of researchers at the University of Kansas now think they’re getting close to cracking the optical clock recovery challenge with a development that would retime not just a single wavelength but a group of up to 10 wavelengths at a time in a DWDM system.

"If we're lucky, this could be a breakthrough as big as the EDFA [erbium-doped fiber amplifier] was," says Kenneth Demarest, a director at the university's Lightwave Communication Systems Laboratory.

To put this into perspective, plenty of labs have found ways of retiming a single wavelength, but very few have managed to retime multiple wavelengths simultaneously. The closest is probably Fujitsu Laboratories Ltd., which demonstrated multiwavelength clock recovery at the 2000 European Conference on Optical Communication (ECOC), about a year after the Kansas team published its theories in the Institute of Electrical and Electronics Engineers Inc. (IEEE)'s Photonics Technology Letters.

The Kansas method exploits a phenomenon called stimulated Brillouin scattering, in which two signals, very close in frequency, travel in opposite directions around a fiber loop. Their combined presence creates an acoustic wave that filters out everything but the strongest peaks. Ideally, those peaks will be the pulses that represent clock beats.

"What you want to do is throw away everything else except the clock tones. This Brillouin phenomenon gives you selective amplification" to let you do that, Demarest says.

The project involved building a module containing such a fiber loop. The idea is that this module could be integrated with a "2R" subsystem, taking the place of the OEO (optical-electrical-optical) retimers that normally accompany those subsystems. A small piece of the optical signal would be split off to the module, which would then work its magic and produce the necessary clock signals.

Brillouin-based clock recovery had been demonstrated before, but not for multiple wavelengths at once. Demarest and his colleagues got the technique to work on two wavelengths simultaneously, and they believe they can extend the concept to more wavelengths, operating on all of them at once the way an EDFA does.

For a long time, the researchers' work was funded by Sprint Corp. (NYSE: FON), but that came to an end with the collapse of the telecom bubble. "We ran out of money and time," Demarest says.

It's easy to see why, as the industry still hasn't reached a point where optical 3R regeneration is all that imperative. Techniques such as forward error correction help avoid the need for regeneration in many cases. And where 3R regeneration is used, the cost of OEO conversions has fallen to the point where it's not a burden at the 10-Gbit/s level, says Dan Al-Salameh, director of optical research at JDS Uniphase Corp. (Nasdaq: JDSU; Toronto: JDU).

"The potential for optical 3R would be more advantageous at 40-Gbit/s and higher, because OEO is still expensive there," he says. But even then, multiwavelength 3R might not be that useful. For example: "If you mix 40 and 10 Gig in the same network, you may need to regenerate 40 Gig more often. Then individual channel regeneration becomes more commonplace," Al-Salameh says.

The lab's technology needs some fine-tuning before being unleashed on the world anyway. The main question is amplification, because the signal needs to be strong enough for the clock pulses to survive the Brillouin filtering. Obviously, one way around this is to use a more powerful amplifier, so Demarest is always looking out for newer, higher-powered amps.

It would also help to use high-nonlinearity fiber, which would introduce (you guessed it) nonlinearities that could, in a sense, magnify the amplification. Along those lines, Demarest has secured a grant with the Photonics Technology Access Program, a National Science Foundation (NSF)-sponsored entity that funds development of specialty devices that target academic research.

"I'd never heard of this until a colleague of mine got an announcement saying [PTAP was] looking for proposals for uses of high-nonlinearity fiber -- fibers with nonlinearities hundreds of times greater than normal " Demarest says. Bingo! Demarest's proposal was accepted, and he's awaiting receipt of the fiber.

The fibers from PTAP, along with the availability of EDFAs stronger than previously available (20dB was the best they could get during the first experiments) give hope that a subsystem handling more than 10 wavelengths is possible, Demarest says.

— Craig Matsumoto, Senior Editor, Light Reading

Be the first to post a comment regarding this story.
Sign In