DEFEATING Dispersion

PART II To reach 40 Gb/s, carriers and vendors need to get a very tight grip on chromatic dispersion

Unlike polarization mode dispersion, which only began "raising its ugly head" with the debut of OC-192 (10 Gb/s) DWDM systems, chromatic dispersion has been an issue that optical engineers have had to wrestle with in network design since Day 1.

In fact, chromatic dispersion is one reason why OC-48 (2.5 Gb/s) DWDM was so necessary and so popular for as long as it was, says Stephen Alexander, vice president and chief technology officer for Ciena.

"Some of the business we got five or six years ago for our 2.5 Gb/s systems was because carriers that were deploying 10 Gb/s systems couldn't get them to work because of chromatic dispersion," he explains.

Will history repeat itself as OC-768 makes its way to commercial availability? In the second part of this two-part series on dispersion (Telephony, Nov. 27, page 90), we'll take a closer look at chromatic dispersion and the steps vendors and carriers have taken to alleviate its effects as they deploy longer spans and faster systems.

It stands corrected Chromatic dispersion is introduced when a signal is modulated onto a fiber. When the light source that launches the pulses that represent ones and zeros is turned on and off, it clips the edges of the signal, causing some wavelengths in that signal to become shorter than others. As a result, the shorter wavelengths travel faster down the fiber than the other wavelengths that make up the signal, causing the pulse to spread.

Ironically, some chromatic dispersion is necessary because it helps to prevent another effect called four-wave mixing, which occurs when wavelengths spread out enough to begin interfering with one another. So because some chromatic dispersion is desirable, it always will need to be corrected.

Although single-mode fiber has no dispersion in the 1510 nanometer window, it has quite a bit in the 1550 nanometer region, which is where DWDM systems are built because of the low loss in that window, says Mark Barratt, vice president of business development for LaserComm, a company developing products to alleviate the effects of chromatic dispersion.

It takes five times the amount of dispersion compensation to correct chromatic dispersion on single-mode fiber, he adds. New non-zero dispersion shifted fiber (NZDSF) has been manufactured to produce less noise and less dispersion to take signals longer distances, but it still requires chromatic dispersion correction nonetheless.

In addition to hampering carriers with older fiber in their networks, chromatic dispersion also has been a challenge to those building and deploying ultra long-haul DWDM systems. At 10 Gb/s, light pulses launched by standard modulation techniques only travel about 400 kilometers on standard single-mode fiber to 600 kilometers on NZDSF before special chromatic dispersion techniques are required, Barratt says. It is important to note that ultra long-haul system vendors such as Corvis and Qtera, which was bought by Nortel Networks, are skirting the issue by using high-quality, low-noise RAMAN amplifiers, different modulation schemes and inverse multiplexing of 2.5 Gb/s channels to reach 10 Gb/s capacity at ultra long-haul distances, according to analysts and vendors.

To reach the distances they reach today, carriers and vendors deploying standard DWDM systems use a component known as dispersion compensating fiber. Dispersion compensating fiber is a spool of fiber several meters long that has fixed values of dispersion, which cancel out the fixed values of dispersion on the carrier fiber. The dispersion compensating fiber is placed inside an amplifier where it cancels chromatic dispersion's effects as a light pulse on multiple channels passes through it.

Another fiber-based chromatic dispersion correction technique - fiber gratings - reflects each wavelength individually, slowing the shorter ones down and allowing the longer ones to catch up. Unfortunately, chromatic dispersion-correcting fiber gratings have been difficult to make and mass-produce, says John Lively, senior analyst of optical components for RHK. However, they may yet have some unrealized potential, as 3M is pursuing their further development, he adds.

Nevertheless, it is likely that it will take more than dispersion compensating fiber or fiber gratings to correct chromatic dispersion in future networks, Barratt explains.

A slippery slope Today, vendors use dispersion compensating fiber to simultaneously correct the chromatic dispersion on every channel being transmitted across the C-band of erbium doped amplifiers (1530 to 1565 nanometers). However, because dispersion compensating fiber initially was created to diminish chromatic dispersion on a single wavelength, when it is used on a fiber carrying multiple wavelengths, it typically is set to correct chromatic dispersion most accurately on the center wavelength.

Because each channel needs a different degree of correction, channels positioned farthest away from the center wavelength receive less accurate correction than those closest to the center wavelength. This property is called dispersion slope.

Correcting for dispersion slope may or not be necessary. But it has become a bigger factor as spans have lengthened, data rates have increased and as carriers have begun to pack more channels onto a single fiber. To correct chromatic dispersion more precisely, some carriers and vendors use multiple loops of dispersion compensating fiber to correct small groups of channels across the C-band.

Continuous slope matching correction will be required to deploy OC-768 DWDM systems and to facilitate longer span lengths of ultra long-haul systems, especially as transmission expands into the L-band (1570 to 1610 nanometers), Barratt says. Because correction values differ according to fiber types, different chromatic dispersion devices are needed for each fiber type, he adds.

"Life gets pretty bad at 40 Gb/s," Barratt says. "Chromatic dispersion goes up by the square of the bit rate. It's 256 times worse at 40 Gb/s than it is at 2.5 Gb/s. Once you get to 40 Gb/s, you need to solve chromatic dispersion very precisely and PMD has to be resolved as well."

Upper management LaserComm has designed a line of dispersion management devices it claims compensate for dispersion and provide continuous slope matching correction. Currently in beta testing with customers, LaserComm's Hi-Mode dispersion management devices are slated for commercial availability in the first quarter of 2001, Barratt says.

LaserComm estimates the Hi-Mode dispersion management device will save users 40% of the cost they would pay to go longer distances without it because it will enable regenerators to be spaced farther apart. Half of the savings will come from Hi-Mode's ability to tolerate more power than dispersion compensating fiber, which Barratt says will allow vendors to use simpler, less expensive amplifiers and enable new features such as an add/drop capability.

The Hi-Mode dispersion management device works by transforming the light from standard mode to high-order mode. "Once you do that, the physics change and you can do different things that you can't do to a signal in standard mode," says Darrell Maronde, director of marketing for LaserComm. "It's kind of like going from two dimensions to three dimensions. You can get around things that would normally block your path."

Another new entrant to the space, Avanex, has introduced its PowerShaper line of dispersion management processors.

The product portfolio includes a processor for 40 Gb/s applications, which features a wide passband and adjustable dispersion slope compensation; a variable compensator for use in the C- or L-band; the PowerShaper8, a 23-inch rackmount shelf that houses eight units and is compatible with all fiber types; and its Super Power Shaper, a variable high-order dispersion and slope compensating processor.

Avanex's technology is based on a patent the company licensed from Fujitsu for what they call a virtually imaged phased array device, RHK's Lively says.

In addition, earlier this year, Corning introduced new PureForm dispersion compensation modules (DCMs), which correct for slope in the company's single-mode fiber and LEAF (large reflective area fiber).

Carriers hope that vendors will do whatever they can to make their DWDM systems as robust as possible, so they don't have to deal with "fine-tuning anything," says Tom Afferton, district manager with responsibility for advanced transport technology and architecture planning for AT&T.

While some vendors are content with the performance and continued improvements of dispersion compensation fiber (DCF), others "see a problem going ahead," says Dana Cooperson, director of optical transport for RHK. Equipment manufacturers of ultra long-haul DWDM systems already use DCF that specifically corrects small groups of channels that make up a fraction of the C-band. Some vendors think they will need to switch to individual channel correction to go as far and fast as they want to go, she adds.

When slope correcting chromatic dispersion compensators are available at the right price, Ciena will switch from engineering around the problem and using multiple loops of DCF to correct small groups of wavelengths to the new alternative, Alexander says.

"Either way, we are expecting service providers to spend more money on chromatic dispersion than they have in the past," Cooperson adds. "They are looking at it as a trade-off. Spend more on chromatic dispersion, but keep signals in the optical domain longer and therefore spend less on electronics." The worldwide market for chromatic dispersion compensation devices is expected to reach $341 million in 2001, $504 million in 2002, $704 million in 2003 and $837 million in 2004, according to RHK.

"The market for DCMs is going to grow rapidly in the near future, driven by the advent of ultra long-haul and 40 Gb/s systems," Lively says. "Both applications require greater numbers of DCMs per link and higher priced, higher performance devices having 100% slope compensation, and in some cases, dynamically tunable slope compensation. Overall, we are forecasting a 47% compound annual growth rate for this market through 2004."

ElectroniCast has issued a market forecast for grating- or filter-based package DCM, which predicts that the market in North America will grow from 17.13 million in 1999 to $115.78 million in 2004, reaching as high as $176.34 million in 2009.

Because tight chromatic dispersion correction is needed for ultra long-haul systems, it affects both 2.5 and 10 Gb/s systems today. And at 40 Gb/s, there is a general industry feeling that chromatic dispersion may have to be corrected for each wavelength or a small group of wavelengths and be done dynamically as well, Cooperson says.

"Whether or not we will be able to get by without dynamic chromatic dispersion compensation will depend on many factors. I think it's still too early to tell," says Rod Alferness, chief technical officer for Lucent Technologies' optical networking group.

"The margins over what is the `right' compensation will be so short that it will be very difficult to design systems and stock the multitude of DCF or other compensating tools with fixed devices that folks will expect dynamic compensation tools," Cooperson says. "Several vendors are already working on this approach."

Compensation? People in the vendor community are interested in whatever anyone has to show regarding chromatic dispersion compensation, but few seem to believe a working solution for chromatic dispersion at 40 Gb/s is "at hand" today, says Kevin Slocum, managing director for Wit SoundView.

"I think what LaserComm has looks interesting, but it is still a little bit early," he adds. "As you know in this market, however, the biggest issue today does not seem to be getting the order. If you prove you have a solution, the next thing you have to do is prove you can make lots of them. Before [vendors of dispersion compensation devices] can have success, they have to prove they can do both."

There definitely is a market for new chromatic dispersion solutions, says RHK's Lively.

"I think it is still too early to declare technology winners and losers [in the chromatic dispersion market]," Lively says. "What is clear is that conventional fiber-based DCMs have not yet been able to provide all the desired attributes of DCMs for ultra long-haul and 40 Gb/s systems such as 100% slope correction and tunability. This has created a real market window for a company that can find a way to do this."