By KATHLEEN RICHARDS
The quest for bandwidth beyond the C-band is the driving force behind the L-band products introduced in the last year. As some carriers begin testing and, in a few cases, even deploying 64-wavelength systems, new developments in short-band (S-band) technology also point toward commercial products. Yet in today's more constrained environment, is the market ready to consider the S-band as another avenue to greater capacity in communications networks?
Startup Xtera Communications (Allen, TX) is one company that is banking on it. Xtera got the industry's attention at the Conference on Optical Fiber Communications (OFC) in March by demonstrating a DWDM subsystem that uses dual-stage discrete Raman amplification-the gain medium is within the amplifiers. A post-deadline OFC submission detailed an experiment in which Xtera was able to get 20 channels in the S-band working at 10.7 Gbits/sec over a 10x85-km span on singlemode fiber without forward error correction, with 14-dBm output power from cascaded Raman amplifiers. The use of higher amplification (>19 dBm) and forward error correction would result in 80 S-band channels, according to Xtera.
Called the eXpander System Integral, the Xtera subsystem features amplifiers, add/drop multiplexing/demultiplexing, coupling, monitoring, and management for the S-band wavelengths in a single shelf of equipment. Outside of the coupler, other companies supply the components. An OFC demonstration of multiband transmission (C- and S-band) used S-band 40-channel arrayed waveguide grating multiplexing/demultiplexing components from Lightwave Microsystems and passive Raman amplification technology from Oplink Communications.
"Through the pump selection and the gain media that we are using, we are actually able to achieve a discrete Raman amplification in a package that is palatable to our customers in terms of size, thermal management, power levels, and power consumption levels," says Mark Nietubyc, technical marketing manager at Xtera. "We are also able to do a dispersion compensation of the S-band wavelengths within the amplifier so that it eliminates the need for an external DCU [dispersion compensation unit] to adjust for compensation across the span distance."
The company's business model is to sell the shelf to DWDM systems vendors and work with them on the final system integration. Vendors are showing some hesitancy, however. "Right now, it is touch and go, because it is such a new business model," says Nietubyc. "We're in trials with some of our customers at sites to essentially prove-in the system concept with their equipment." Xtera hopes to announce a major system vendor and carrier customer this quarter.
"I certainly welcome Xtera's product; it is a forward-looking technology," says Vladimir Kozlov, senior analyst at RHK Inc. (San Francisco). But Kozlov believes the industry is at least two or three years away from deployment of these systems. The S-band vendors need to prove that their technology works-and carriers have yet to exhaust the capacity available in the C-band (1,530-1,565 nm) or the L-band (1,570-1,610 nm). In addition, technical issues remain, particularly with regard to the amplification technologies and transmission fiber. Erbium-doped fiber amplifiers (EDFAs) do not adequately amplify S-band wavelengths, for example.
Fiber manufacturers have recognized the potential market for fiber that supports S-band transmission, as evidenced by Lucent's All-Wave introduced last year, and Corning's SMF-28e. Companies in Japan have also done a lot of work in S-band transmission, mostly in the area of thulium-doped fiber amplifiers (TDFAs). "A lot of their infrastructure is dispersion-shifted fiber and that puts your zero point smack dab in the middle of C-band, so you end up losing a lot of capacity in terms of wavelengths when you try to do DWDM. So they're looking for solutions that will take wavelengths outside the C-band," says Nietubyc.
Meanwhile, L-band transmission is not without its own headaches. "A lot of carriers are unaware of the absoluteness of some of the physics that they are up against, given a choice of expanding into an L-band or an S-band," says Nietubyc. "There are higher losses associated in the L-band in terms of macrobending or microbending areas of the fiber that are not experienced in the S-band; and on standard fiber, a lower dispersion [is] associated with S-band wavelengths.
"If a route is at the end of its span budget in terms of all the margin that is built-in for the power level used to support C-band transmission, and add L-band wavelengths, Raman backscattering is actually going to work against you because your C-band wavelengths will start losing power to your L-band wavelengths," he continues. "With S-band, you don't experience that only because the S-band wavelengths are lower, so essentially you may actually gain some of that margin back because your S-band wavelengths will pump the C-band wavelengths."
"If they're right about the use of the L-band, that is going to be a rude awakening to carriers," says Dave Krozier, senior analyst at RHK Inc.
One factor in getting carriers to invest in S-band equipment is ensuring a steady flow of equipment. The fact that component manufacturers are as much on the fence about the technology as service providers also impedes action in this area. "There isn't a lot of S-band componentry out there being announced by major providers," Nietubyc admits. "There is a bit of hesitancy on their part-the chicken and the egg syndrome-they don't want to be the first ones out there because they don't want to extend a lot of resources and effort, but at the same time, they realize that it is coming and they want to be prepared-but at what point do they bite the bullet and make the investment?"
Passive device manufacturers are starting to get requests from their customers to provide S-band specifications, however. Several companies, including Agilent Technologies and ThorLabs, offer S-band laboratory and test instruments to support development.
"We're laying a little bit of a bet in terms of what the future holds," says Alex Cable, president of ThorLabs (Newton, NJ). "Initially, the bandwidth requirements for powering the Internet had us feeling as though the C- and L-band would be used up by the '03 time frame. Now we're seeing 2x per year growth in terms of bandwidth demand; I hear some numbers 1.5, 1.7-whatever it is, it is significantly lower than what we all thought was going to happen back in the '99 and '98 timeframe.
"Originally, we worked on the assumption that there was a big unknown," continues Cable. "In talking to people in the industry, we came to the conclusion that what was needed today was tools for the R&D laboratories and tools for the component manufacturers that would at least allow them to design products that were compatible with a future push into the S-band."
ThorLabs is offering a TDFA that covers the wavelength range of 1,455-1,485 nm, designed for laboratory applications such as component testing and optical power meter calibration. The device was designed and manufactured in Japan through the company's partnership with Fiberlabs. It uses thulium-doped fluoride fiber pumped with a single laser diode for a gain of about 25 dB with a noise figure of 7 dB and an amplified output power of 13 dBm. The output fiber is SMF-28 silica fiber. ThorLabs also offers an S-band ASC test source.
"There are a lot of challenges that we're faced with in terms of how much coverage is really required in the S-band," says Cable. "The definition of the C-band was made around the erbium-doped silica fiber EDFAs. The S-band in my mind is a little more arbitrary. We have a TDFA made with a fluoride fiber and the window doesn't cover the whole S-band. So what we're working on is how much of the band needs to be covered in a commercial instrument."
Cable reports that the company's S-band amplifier currently has about a 30-nm window. A move to telluride glass would multiply that figure by at least a factor of 1.5, he feels. However, such a move probably would require another 18 months to complete the development work.
NEC Corp. (Nakahara-ku, Kawasaki, Japan) is also doing work with TDFAs. "We have developed them for our lab use only," says Tadashi Kasamatsu, assistant manager at NEC. "From a practical standpoint, they have no severe essential problems. However, what I should note is the reliability of thulium fibers, the gain fiber. Thulium fibers are based on fluoride glass, which is still considered to be in engineering samples. We expect improvements in thulium fibers by the fiber manufacturers."
Another issue is Raman scattering in link fibers, Kasamatsu reports. This effect degrades the optical signal-to-noise ratio of S-band and gives unequal signal-to-noise ratios. This effect is partially canceled by distributed Raman amplification in the link fiber, but not perfectly-the distributed Raman amplification generates additional noise.
"I cannot say the exact timing of when S-band systems will be deployed," says Kasamatsu. "Since the situation is changing very fast and many promising candidates are competing with each other. However, I do not think that they will predominantly be installed in Japan."
"I feel that it is still early in terms of deployment," says Cable. "My guess is that '03 isn't going to happen, that there are too many financial constraints facing the industry and that the recovery is going to push us off well into '04. But certainly, we are excited by the growth that we're seeing in the demand for the testing labs."
Cable remains confident that lab interest will eventually translate into commercial equipment. "It is much easier to extend the number of channels than it is to increase the modulation bandwidth, and that's why it's inevitable that the S-band will be populated," he concludes.