Scientific American article which discusses strides in all-optical techology. Courtesy of pass pass on the CIEN Board:
sciam.com
Enjoy. Comments Welcome.
-------
"NOTHING BUT LIGHT"
Hunger for bandwidth drives all-optical technology to market
The Internet-fueled boom in data communications has set off a grab for bandwidth--the additional network capacity needed to transmit Monica Lewinsky's grand jury testimony or the Taliban's Web page. Traffic on the Internet as much as quadruples every year, whereas plain old voice calls chug along at 8 to 13 percent annual growth. To sate the bandwidth crunch, long-distance telecommunications carriers have begun to demand optical communications technologies that had languished in university and industrial laboratories until the mid-1990s. "There's a useful place for the technology to go," notes Steve W. Chaddick, a senior vice president at Maryland-based Ciena, a leading optical network equipment manufacturer. "That wasn't true just a few years back."
Five years ago networks that incorporated what is called a dense wavelength division multiplexer (DWDM) were to be found in U.S. and European government-industry research consortia that were showcasing new technologies. This heavy-handed engineering term describes networking equipment that has, in the interim, rescued long-distance carriers such as the telecommunications provider Sprint from a bandwidth drought. The multiplexer sends laser light of different wavelengths down a single optical fiber. Meanwhile components of the transmission system in the path of the fiber reflect individual information-carrying wavelengths, allowing them to be diverted onto or off a high-capacity link. DWDM systems work in concert with optical amplifiers that can boost the strength of many wavelengths at once without having to convert the wave back into an electrical signal.
With this technology, the capacity of in-the-ground fiber can be expanded by simply adding wavelengths. For Sprint, deploying the multiplexers costs roughly 40 percent of the $77,000-per-mile expense of adding new fiber. "We would have had serious problems without this technology," remarks Frederick J. Harris, Sprint's director of network planning and design, whose company uses DWDM on 90 percent of its 30,000 miles of fiber networks.
The U.S. market for this technology grew from nothing in 1994 to $1.5 billion last year and is expected to reach more than $4 billion in size by 2001. "Supply for bandwidth still has not crossed demand, so the market for the technology continues to grow," says Mathew H. Steinberg of the market analysis firm RHK in South San Francisco. (Before 1994, a small market existed for wavelength division multiplexers with only two channels.)
To meet new growth, multiplexers will flirt with or break the terabit (trillion-bit) barrier on a single fiber; a trillion bits per second exceeds all the traffic on the Internet. Most current equipment tops out at about a tenth of a terabit. But several firms--including Pirelli Cables and Systems North America in Lexington, S.C., and Lucent Technologies in Murray Hill, N.J., as well as Ciena--are either shipping or readying delivery of equipment that can support from 80 to 128 wavelengths on a fiber, each wavelength carrying up to 10 gigabits of information. Lucent Technologies's Bell Laboratories will attempt an experiment next year that would transmit 1,000 wavelengths on a fiber, in an effort to test the maximum capacity an individual fiber can accommodate.
Multiplexers create the lanes on optical superhighways. But these pathways move only from point A to point B. To channel traffic from New York to either Los Angeles or Seattle, a switching interchange may be needed in Chicago. So companies have dusted off 1980s-era research on switching optical signals.
Light-wave switches would avoid the costly burden telecommunications carriers now face--converting the multiple gigabit stream running on each wavelength into dozens or hundreds of lower-speed electronic signals, switching them and then reaggregating them into a single light channel. Huge telecommunications equipment companies and start-ups alike are now racing to develop all-optical switching products. Photonics has even become a basis for regional economic development. In late October a group that combines the University of Texas at Dallas, several venture capitalists and major telecommunications equipment suppliers and carriers announced the establishment of a photonics development center based in Richardson, Tex., intended to attract new companies to the region.
Optical switching elements, expected in 1999, will be incorporated into the next generation of DWDM products. They will allow any wavelength in a fiber to be diverted onto or off a network on command, unlike current multiplexers, which cannot be reconfigured without a technician first disabling a fiber circuit. Tellium, a New Jersey start-up that was spun off from Bellcore, the former research arm of the regional phone companies, is one of several firms laboring on the technology. It has developed an optical switching multiplexer that uses the polarization state of liquid crystals to add or drop up to 64 wavelengths from a fiber.
Telecommunications suppliers such as Sprint and MCI want more than a souped-up multiplexer. They hanker for the photonic equivalent of an electronic switch called a digital cross-connect, which switches hundreds of incoming signals to an equal number of outgoing channels. Today's digital cross-connects, however, require that the multigigabit light waves that are channeled along fiber networks be converted to lower-rate electronic signals.
MCI Worldcom in Jackson, Miss., has purchased an early version of an optical cross-connect switch to protect against "backhoe losses": the catastrophic curtailment of phone service that occurs when a fiber is cut. The 24 deployed switches, which were manufactured by Astarté Fiber Networks in Boulder, Colo., use a piezoelectric material that steers the light from any of 72 incoming to any of 72 outgoing fibers. This system allows immediate restoration of service if a fiber goes down.
A hand-me-down from a technology used in classified military networks, the switch is very much a first-generation product. Astarté and others are working on switching elements for optical cross-connects that will provide more capacity and reduce the cost and size of the products. Some companies are considering arrays of thousands of microscopic mirrors that can tilt individually to send a wavelength down a chosen pathway. Alternatively, an electric field applied to certain materials may change the way light is routed. With yet another approach, called thermo-optics, application of heat to a polymer blocks light from proceeding down one pathway but not another. "In the next couple of years, you're going to see a shoot-out, and some practical devices will come out of this competition," notes Alastair M. Glass, director of photonics research at Lucent.
Despite the photonic revival, the difficulties of switching signals optically have caused some companies to opt for the development of new electronic switches that can accommodate high-bandwidth pipes. And even if optical cross-connects become ubiquitous, telecommunications specialists see a continuing role for electrons, which may be needed to reshape light pulses that have attenuated over long distances and in monitoring networks. "There's no way anyone knows to determine optically the number of bits with errors on an all-optical signal," says Tellium chief technology officer Charles A. Brackett.
The prospect of terabit networking, however, has begun to prompt further rethinking of how networks operate. In the laboratory, work continues on the speculative idea of switching not just wavelengths of light but the individual packets of data transmitted over fiber networks, all of which are now processed with relatively slow electronic switches. A European consortium, ACTS, has demonstrated an optical router that performs this function. "This type of device might handle routing and forwarding of data with multiple terabit inputs without slowing down traffic," says Daniel J. Blumenthal, associate professor of electrical and computer engineering at the University of California at Santa Barbara. Blumenthal is attempting to build a prototype optical router that forwards packets using the Internet Protocol.
For the moment, optical packet switching is still a dream. But the pull from a marketplace that is warming to the idea of a trillion bits per second may help turn laboratory oddities into commercial realities.
--Gary Stix |
