As OC-192 (10 Gbits/second) reaches mainstream status, many new companies are turning their eyes toward the sheer physical difficulty of OC-768 (38.8 Gbits/s). Meanwhile, systems companies are developing all-optical switching schemes, providing competition for the likes of Corvis and Lucent Technologies Inc. And a number of companies are vying to realize the elusive dream of using polymer materials to integrate photonics components, a step that would be analogous to the creation of ICs in electronics.

The OC-768 speed grade, generally expected to reach volume production in 2003 or earlier, will pose new physical challenges at the photonics level.

Among the problems at 40 Gbits/s is chromatic mode dispersion (CMD), or the spreading out of a light pulse as it travels long distances. The danger is that too much dispersion will blur the difference between ones and zeros, destroying the signal.

The so-called positive dispersion present in the fiber can be counteracted by introducing a length of fiber with the corresponding "negative" dispersion; the two cancel each other out. But that practice falls apart when more than one wavelength is being transmitted, because each wavelength has a different rate of dispersion, said Mark Stubbe, vice president of marketing for LaserComm Inc. (Plano, Texas).

LaserComm uses a physics trick to compensate for CMD in each wavelength of a wavelength-division multiplex transmission. The company's device, called Hi-Mode, converts each wavelength to a higher harmonic mode, at which it is possible to compensate precisely for each wavelength's dispersion. The procedure has drawn attention since the early '90s, Stubbe said, but LaserComm claims to be the first to have perfected the technique.

LaserComm already has C-band versions of the product in trials with DWDM suppliers and plans demos of both C- and L-band versions at OFC. Those products, due for volume shipment in the second quarter, are passive devices aimed at OC-192 speeds.

For OC-768, LaserComm plans an active device that will use feedback loops to detect any drift in the compensation process, which becomes more of a threat at higher speeds. "If the fiber temperature changes from morning to afternoon, that's enough to cause bit errors [at OC-768]," Stubbe said.

Another startup, Phaethon Communications (Fremont, Calif.), will use OFC to unveil a dispersion management device based on a fiber Bragg grating. A small, electrically driven engine stretches the grating by up to 0.1 percent, "enough to create additional delay in the signal or reduce delay in the signal," chief executive Gary Cuccio said. The device uses a feedback loop to monitor the optical signal and adjust the grating as necessary.

Another problem at 40 Gbits/s is polarization mode dispersion (PMD), in which the light wave's polarity drifts. Startups such as Yafo Networks (Hanover, Md.) are developing ways to counteract PMD. Phaethon is testing three methods and hopes to combine one of them with its CMD device eventually, Cuccio said.

Given the physical complications facing OC-768, it's likely that the speed grade will not debut in the network core. Rather, something akin to the very short reach (VSR) standards — using 10-Gbit/s transmission between boxes in a central office — may prove the launchpad for 40-Gbit/s signaling.

To see OC-768 debut in short-range applications "wouldn't surprise me at all," said Steve Alexander, chief technical officer of Ciena Corp. (Linthicum, Md.). "There's been an inversion of where people expect higher rates to show up."

All-optical switching

While true all-optical networking still awaits advances such as optical memory, companies that use all-optical switch fabrics are emerging. At least two startups — Ilotron Ltd. (Kent, England) and Luxcore Networks Inc. — will unveil their plans for all-optical switches at OFC.

Aside from Corvis, which claims all-optical capability but won't disclose its technology, most of the publicity surrounding pure photonic switching has gone to microelectromechanical systems (MEMS), in which tiny mirrors are adjusted to switch light to the proper ports. Several startups are developing MEMS switch fabrics, and systems from Lucent, Luxcore and Ilotron will combine those switch fabrics with routing and provisioning.

Ilotron uses the MEMS fabric from Optical Micro Machines Inc. (OMM; San Diego). Ilotron found OMM's two-dimensional MEMS structure more practical than three-dimensional, rotating-mirror components, said Scott White, vice president of business development. "We're using 2-D MEMS because when you come down to the crunch, that's all that's commercially available today."

Luxcore officials wouldn't say which MEMS switching fabric their company is using.

Both companies claim to have solved a primary hurdle to all-optical networking: all-optical wavelength conversion. Luxcore will announce a proprietary tunable module for automatically converting wavelengths. The company uses an out-of-band signaling scheme to provision circuits between its boxes and prevent contention for wavelengths, said Calvin Martin, vice president of optical systems engineering.

Ilotron plans to handle wavelength conversion by using tunable transponders. Future versions of the company's switch will be able to drop wavelengths into the transponder for conversion, White said. The process is working in the lab but isn't commercial-grade yet, he said.

Lucent, meanwhile, will use OFC to tout its LambdaRouter all-optical switch, which began shipping in July and is being used in trials by Global Crossing Ltd. (Hamilton, Bermuda). The switch can handle 256 input and output ports, and a 1,034-port version is being readied for release possibly by year's end, said Eric Spurrier, Lucent's vice president of marketing for optical networking.

For years, the photonics industry has been working to develop products based on polymers — plastics doped with particular molecules to yield a high electro-optical coefficient. Telephotonics is one of those developing products based on homegrown polymer technology. It will display its first products at OFC.

The promise of polymers lies in the ability to integrate components, as is done in CMOS. That could be particularly important for OC-768, where the modulator driver would be a costly 4-V indium phosphide semiconductor, he said.

Polymer matters

In fact, the greatest future for polymer may be as an interconnect technology, where it would be used to link optical components on one substrate.

Pacific Wave Industries Inc. (Los Angeles) will unveil its efforts to build a 40-Gbit/s modulator using an electro-optical polymer material it calls PCCOD. A modulator creates zeroes and ones by blocking a laser signal or allowing it to pass through (that technique is required at higher speeds, where blinking the laser on and off might affect system reliability).

Pacific Wave is starting with a 40-Gbit/s modulator but hopes to move to a 10-Gbit/s product to compete with existing modulators built of lithium niobate (LiNO₃). Company officials say the key to their products is the ability to handle such properties as the thermal stability of the material, which was developed in research efforts at the University of California, Los Angeles, and the University of Southern California.

Early sales of the Pacific Wave OC-768 modulator will be to universities and research labs. Compliance with Telcordia standards should be completed in six to nine months, the company said.

LiNO₃ modulators for OC-768 are also under development by Lucent, Codeon Corp. (Columbia, Md.) and others.

Gemfire Corp. (Palo Alto, Calif.), meanwhile, is producing what company officials believe is the first polymer material that will meet Telcordia standards for telecommunications equipment. Gemfire's first products, being displayed at OFC, include an eight-channel optical attenuator and an eight-source pump laser for erbium-doped fiber amplifiers. Both show off the company's integration capabilities, chief executive Rick Tompane said.