Technology
By KATHLEEN RICHARDS
Essex Corp. (Columbia, MD), an optical-engineering company working on optical and signal-processing technologies for the defense and U.S. intelligence industries, has developed a 1-GHz channel-spacing technology that it plans to market for last-mile fiber-optic communications.
Dubbed Hyperfine Wave Division Multiplexing (HfWDM), the technology is being developed for passive optical devices such as demultiplexers, multiplexers, add/drop multiplexers, and tunable receivers. The Hyperfine devices will support applications targeted at the small business and fiber-to-the-home markets.
"The 1-GHz channel spacing is basically accomplished in the Hyperfine device; we're really not describing specifically how it works because we have a patent pending," says Terry Turpin, senior vice president and chief technical officer at Essex. "But basically the device is analogous to a Bragg cell in that it produces different angular outputs for each wavelength that goes into it."
A Bragg cell is a block of transparent material with a piezoelectric transducer on one end. "You apply a voltage across the transducer and launch a soundwave down the device," says Turpin. "Then you illuminate the transparent material from the side and the soundwave appears to be a moving diffraction grating. The different frequencies of sound produce different frequencies of diffraction grating and diffract the light in different directions. The technology that we have doesn't involve any acoustic intermediary, but what we are trying to do is basically replace the [Bragg cell] device."
The technology wasn't initially developed for fiber-optic communications applications. "We were working on trying to develop an electronic-warfare receiver," explains Turpin. "These things are built with Bragg cells as the electronic-to-optical transducer and they typically have 1 to 2 GHz in bandwidth and a resolution of about 50 MHz. We had a requirement for a device that had 16 GHz of bandwidth and 50 MHz of resolution. We finally came up with an approach and realized that it was a very flexible technology that had virtually unlimited bandwidth and could produce channel spacings anywhere from 50 GHz down to 50 MHz. We can also very effectively shape the passbands of those channels, so instead of having the normal (Sin(x))/x passband, we can have nice flattop passbands."
Turpin sees a ready market for the technology in the last mile of the network: "We are currently targeting passbands on the order of 1 to 2 GHz, and the reason for that is that we are looking at the small-business markets, and perhaps fiber-to-the-home, where we would like to have lots of very fine channels to be able to give each user a wavelength. One gigahertz nominally fits the lower-end SONET technologies like OC-12. It also accommodates Gigabit Ethernet, which is where a lot of people feel the future really is."
According to the company, the technology is very low cost, an important consideration in the home market. "Instead of building a costly terminal where you pack things together and send them at high data rates and unpack them, we are just going parallel with lots of channels at low data rates," says Turpin. "You can stack these channels up as you need to, to carry your high-data-rate channels. The individual modules are very inexpensive. Another reason that we selected the band around 1 GHz is that it is as fast as you can go and stay in all-silicon CMOS technology, which allows you to keep the cost of the technology down."
Turpin expects the technology will have wide application, including programmable add/drop multiplexers, tunable add/drop multiplexers, and tunable receivers. He predicts that the focus on very-low-cost, small-business and home applications may result in a low-cost box that sits on top of a television set.
Last November, Essex demonstrated Hyperfine technology in a system that supported 1-GHz channel spacing at 16 GHz in the lab. The company's plan is to demonstrate a system that supports 100 and 50 channels with 1-GHz spacing this year. Essex is working closely with potential customers.
Dispersion and nonlinearities do not seem to be an issue, according to Turpin. "Our individual channels are so narrow that if you send everything in parallel, the bits from the upper channels and lower channels may get skewed, but they don't get distorted because of dispersion. We're running tests on nonlinearities and the initial results indicate that we don't have a problem there either."
Essex got the opportunity to work on devices for optical communications when it received $2 million in private funding last September from Networking Ventures LLC and the Global Environment Fund. Founded in 1969 as a human-factors company, Essex acquired an optical and signal-processing company, System Engineering and Development Corp., in 1989. In addition to its defense work for the U.S. government, Essex had a team of engineers working with Motorola on Iridium, the space-based cellular communications system. Today, the company has a core team of 17 engineers.
"The investment permitted us to develop the technology that we had been thinking about but didn't have the resources to pursue," says Len Moodispaw, president and CEO. "In the last two or three months, we have concentrated on developing Hyperfine and some other technologies where the optical processing and the communications come together."
The best use of the technology immediately is the last mile of the metro, where customers don't need large bandwidth and what bandwidth they do need must be fairly flexible. "It doesn't mean that we don't think applications also exist in the long-haul," says Moodispaw. "It's just that there are lots of other people playing in the long-haul and very few people have the answers in the short-haul, we believe."