PRODUCT FEATURE: Optical-switching technology at watershed for high- and low-density applications
By Anita S. Becker, Associate Editor
The vision of all-optical switches is coming into focus for systems designers. This year, liquid-crystal and micro-electromechanical systems (MEMS) -enabling technologies should take major strides out of the research lab and into commercial production for use in optical-switching applications.
Both technologies have the potential of being viable, reliable, cost-effective, and high-performing building blocks for future all-optical transport systems. Yet even at this early stage of their development, the two technologies already appear to be heading for different applications.
Optical switches are widely proclaimed as being the forerunner to the all-photonic network. Positive indicators for the shape of things to come can be found in the behind-the-scenes whispers of upcoming introductions from a variety of companies and the recent product announcements from Corning Inc. (Corning, NY) and Lucent Technologies (Holmdel, NJ). Corning announced its liquid-crystal-based PurePath Wavelength Selective Switch in September 1999 and, soon after, Lucent unveiled the WaveStar LambdaRouter (which is not a router but actually a MEMS-enabled optical crossconnect) in November.
Companies such as Astart‚ Fiber Networks Inc. (Boulder, CO)/Texas Instruments (TI-Dallas), AT&T Labs Research (Red Bank, NJ), and Bell Labs/Lucent, and Cronos are all working on developing MEMS-based optical-switch devices. Chorum Technologies (Richardson, TX), Corning, and SpectraSwitch (Santa Rosa, CA) are working on liquid-crystal-based products. At press time, only Lucent and Corning had made public announcements, although other companies are expected to unveil products in the near future.
Researchers indicate that MEMS and liquid crystal may not compete for use in the same applications. High-density, optical crossconnects may become a reality through the intrinsic scalability and cost-effective, semiconductor-like fabrication of MEMS, a proven technology already being used for motion control in automotive and industrial applications. MEMS technology allows for the building of microscopic mechanical devices onto the surface of a microchip. The use of microscopic, moveable micro-mirrors in optical switches, crossconnects, add/drop multiplexers, and other network equipment offers several advantages over conventional devices, including lower cost, faster speeds and smaller space requirements.
Meanwhile, liquid crystal is being perfected for lower port counts and protection, monitoring, restoration, and routing applications. Liquid crystal exhibits properties in between those of a crystalline solid and a flowing liquid. Its rod-shaped molecules move together as a unit and align in parallel to electric or magnetic fields. Liquid-crystal switches operate by rotating the two perpendicular polarization states of light. Switching between two molecular alignments requires applying voltage. Liquid crystal can be used to switch light, select wavelengths, attenuate signal intensity, and store information.
"If you look at switching applications, they essentially fall into three different categories," explains Cindana Turkatte, vice president of marketing for SpectraSwitch. "You've got large optical crossconnects, which are necessary to replace the digital crossconnects that exist in the network today. There are also a lot of applications for switching in basic fiber protection and wavelength protection-whether it's a four-fiber ring architecture or an optical add/drop multiplexer-where you might want to switch from one channel to the next. I think people have completely underestimated the volume of optical-switching applications for protection and routing, which [use] low port-count switches. That's what's happening in the year 2000."
Turkatte believes the industry wants to move away from mechanical-oriented devices toward nonmechanical, solid-state technology. "Liquid crystal is an excellent platform for a solid-state component, especially in low port counts-meaning anything less than 32 ports, for example, in 1x2, 1x4, or 8x8 switching fabrics. MEMS is going to be possibly the only technology that may prove out for large matrices, 512x512 or 1024x1024 crossconnects. Liquid crystal will never do that as far as we know. Liquid crystal is well suited for where multiple functionality is required-not just for switching but for where you want attenuation or variable coupler functions."
The four-year-old SpectraSwitch is formally announcing its liquid-crystal-based WaveWalker family of optical components at the Optical Fiber Conference (OFC) being held in Baltimore this month. It is also demonstrating a small 1x4 liquid-crystal-based optical switch, according to company president Nick Lawrence. He says SpectraSwitch plans to deliver early samples to its original equipment manufacturer (OEM) customers during the second quarter with production scheduled this summer. The company and other component vendors developing liquid-crystal-based optical devices are meeting at OFC to establish an industry forum as part of a standardization effort.
"What SpectraSwitch and others have done is to recognize the inherent advantages of liquid-crystal technology in working with photons," explains Lawrence. "We are actually leveraging the 25 years or so of display industry experience and adapting the material to telecommunications." The adaptations are in the area of increased performance characteristics. According to Lawrence, past temperature and moisture problems have been solved. "Material sciences have progressed like all the other technologies over the past 15 years. The issue of environment degradation and degradation over time was an early phenomenon because we were working with materials that were either contaminated directly in the processing of the material or became contaminated as moisture migrated through the adhesive barriers that contained the material."
Corning's liquid-crystal-based optical switch, the Wavelength Selective Switch, is part of its PurePath family of products. The company is targeting the new switch for metropolitan and long-haul dense wavelength-division multiplexing (DWDM) systems.
It was introduced with Corning's Pure Path Wavelength Modular Switch at the National Fiber Optic Engineers Conference (NFOEC) in Chicago last September.
Corning's liquid-crystal-based switch can switch up to 40 channels individually, each at 100 GHz, according to the company. System equipment manufacturers can use it to provide dynamic wavelength routing for optical path protection or for the interconnection of metropolitan rings. The switch can also be used as a 40-wavelength, reconfigurable add/drop module. Corning says samples will be available "in the first half of 2000," with a pilot manufacturing line scheduled for the second half of this year.
Chorum Technologies (Richardson, TX), incorporated three years ago, is now developing commercialized optical-networking technology for both telecommunications and data-communications applications. All of Chorum's optical-networking products are based on polarization processing. "It's an approach where you manipulate or control the polarization state of light to perform optical switching, optical filtering, or optical routing and also optical signal conditioning," explains vice president of marketing and business development Doug Dickerson. "You can create a number of different devices using that basic technology approach."
At NFOEC '99 Chorum announced its PolarWave family of products, which includes optical switches enabled by liquid-crystal technology. "What we introduced at NFOEC were a range of optical-switching products, including 1xN and small NxN products," says Dickerson. "We've got a rollout plan that extends from NFOEC last fall to 2000, so every few months we'll be announcing a new product line throughout the year. We have optical switches based on liquid crystal out with customers now."
Dickerson expects production to begin around the middle of this year. "With the liquid-crystal switches, we're introducing products with very good performance comparable to mechanical [devices] but with the reliability of solid-state technology."
MEMS TECHNOLOGY TURNING CORNER
While companies pursuing liquid-crystal technology appear to be focusing on relatively small matrices, MEMS developers have bigger things in mind. "When you say 'all-optical switch,' you are talking about a range of switching. I think what is pulling everyone along [in MEMS development] is what I call the 'mega-switch'- that would be 1,024-port switching," explains Jesko von Windheim, vice president of marketing and business development for Cronos Integrated Microsystems Inc. (Research Triangle Park, NC). "When you look at large switch fabrics-256x256 and up-you realize that around them there is going to be all sorts of other components that eventually will have to be all-optical, too, to really make this thing work. It's kind of like a domino effect-once you get that large switch to work, then you have to start looking at all the other things that feed into and out of it, and you realize there are many, many devices there that must be implemented."
Click here to enlarge image
Cronos recently received venture-capital money, including some from Intel, to develop its own version of a MEMS-based optical crossconnect. "We certainly have strong relationships with a number of companies, but we have not publicly announced [optical-switching] products," says von Windheim. He adds that the company expects to make announcements this quarter. Cronos introduced a sliding mirror during the fourth quarter of last year. "It's applicable to a number of applications, such as variable optical attenuation, simple add/drop [multiplexers], and simple arrays such as 2x2 or 1xN switch arrays. We have a number of customers evaluating those simple switches for those kinds of applications."
Optical crossconnects represent a market waiting for a solution. San Francisco-based Ryan Hankin Kent (RHK) predicts the total DWDM optical crossconnect market to reach $1.8 billion by 2003. "What's to be determined is how much of that is going to be all-optical," says Lynn Hutcheson, an analyst with the research firm. "DWDM is the most near-term application for optical crossconnects. In addition to all-optical products, there are a number of companies working on hybrid optical-electrical-optical (OEO) devices. Those devices are further ahead than the all-optical, but nothing of substance has really been deployed yet. We don't think there's going to be much deployment until 2001."
MEMS is one of the all-optical technologies that Hutcheson is optimistic about. He says that although Lucent was first with its announcement of a MEMS-based optical-switching product, the WaveStar Lambda Router is still under development. "From what I know, I would say that Lucent is ahead [of its competitors], but I wouldn't say they are way, way ahead."
Lucent's Charles Roxlo, director of the advanced-technology department, Optical Networking Group, feels that MEMS is just right for very-high-capacity optical crossconnects. "[Central offices] have more and more fibers coming into them, so they're going to need crossconnects that can handle hundreds and thousands of ports-and MEMS technology can do that. I don't know of another technology that can do that cost-effectively. We've evaluated a number of waveguide technologies and they all suffer from [poor] scalability."
Lucent's optical crossconnect will be a "2N" device based on MEMS research and development by Bell Labs, marketed as MicroStar technology, explains Roxlo: "To make a device with N ports, you need 2N mirrors, and with our MEMS technology, you can manufacture a large number of mirrors on a single substrate easily. It's quite easy to make N pretty large. There are some technologies, including some liquid-crystal technologies, that I would call N2 architectures. In other words, if you want to have a strictly nonblocking crossconnect with N ports, you essentially need N2 individual switches. And as you can imagine, once you get to the larger port counts that becomes quite unwieldy."
TINY ILLUMINATED MIRRORS
The LambdaRouter will rely on tiny micromechanical mirrors positioned so that each is illuminated by a single wavelength. The mirrors are tilted so an individual wavelength can be passed to any of 256 input and output (I/O) fibers. All 256 mirrors are fabricated on less than one square inch of silicon. Lucent states that the system will take in any voice, data, or video signal, and each of the optical crossconnect's 256 channels will support wavelengths at Synchronous Optical Network/Synchronous Digital Hierarchy (SONET/ SDH) or optical carrier/synchronous transfer mode (OC/STM) speeds up to 40 Gbits/sec. The LambdaRouter is slated for field trials in July and should be commercially available this December.
Astart‚ Fiber Networks, in partnership with TI, is also developing products incorporating MEMS technology. TI supplies Astart‚ with its MEMS chip technology and subassembly manufacturing, and Astart‚ handles the switch engineering, design, and integration work. "Our particular MEMS technology is a mirror-based MEMS," says Edward Fontenot, president and CEO. "The mirror-based MEMS is used to redirect or steer a light beam. By steering a light beam which emanates from a transmit fiber...appropriately, we can target the receiving fiber and optically couple the incoming fiber to the outgoing [receiving] fiber. We have developed such an architecture and have a laboratory prototype that has been tested and is performing exceptionally well.
"MEMS allows us to implement what we believe is the best architecture: single-stage switching, free space, a high level of modularity, scalable, futureproof, and very cost-effective. Those in a nutshell are the key things we try to achieve," says Fontenot. "There certainly are competing technologies; some may even be more attractive for a particular market segment. I get e-mails from people all over the world wanting to buy very tiny, very-small-capacity optical switches, and we wouldn't use MEMS technology for a 2x2 switch or a 4x4. There's certainly better technology for that range. But the market that we're going after, the telecommunications service-provider market, means that they have to have a large-capacity product. And as we evaluated the architecture principals that we believe is best for this opportunity and the various technologies that we could conceivably use, we concluded that MEMS technology is in fact the best light-steering technology."
The increasing deployment of DWDM-based fiber networks in North America and Europe is proportionally driving up the demand for optical-network equipment. While almost all switching systems of significant port counts will require the conversion of optical signals to electrical for the near term-some say for the next five years-the imminent emergence of simple optical-switch devices into the marketplace may just be the camel's nose in the tent of the all-optical network.
"The need for optical-switching systems will create a five-year, $31-billion market opportunity for optical-switching systems vendors," according to Pioneer Consulting (Cambridge, MA) in its new market report, "Core Optical Networks: Market Opportunities for Optical Switching Systems in DWDM Networks," (see Market Analysis, page 42). "At the component level, device manufacturers will have significant opportunities to create all-optical-switching fabrics, transparent optical interfaces, and more sophisticated provisioning and management systems."
icd.pennwellnet.com |
