Kalkhoven said Cronos was the most important acquisition JDSU has ever made. It was my general understanding that Cronos was quite a bit farther along in assemblies and modules than their competitors.
http://?www.memsrus.com/cronos/CIMSsvcs.html
Cronos is the only full-service, MEMS facility that offers all 3 key MEMS fabrication processes: bulk micromachining, surface micromachining, and LIGA (high aspect ratio).
We also offer a low-cost vehicle for introducing the MEMS technology to your company's applications. Our Multi-User MEMS Processes, or MUMPs™, provides an accessible, inexpensive proof-of-concept, prototyping function.
This is from the Rockwell press release:
Cronos has created a standardized manufacturing platform for MEMS processing and components consisting of simple building blocks that comprise an application-specific approach to MEMS. This approach further streamlines MEMS integration into optical and wireless communications products significantly reducing time to market for component and system manufacturers.
This America's Network article (posted to the The *NEW* Frank Coluccio Technology Forum) lays out some of the optical switching technology debate in plain English. It doesn't answer all the questions and doesn't address MEMS technology's wireless role but I found it helpful in putting some of the networking pieces together. - peggylynn
Mirrors and smoke: The optical challenge Competition to build the all-optical network has brought photonic switching technology into the light of day, but traditional switch vendors still see a lot of hot air.
By Dan Sweeney
The Holy Grail. The term is invoked almost routinely in reference to the all-optical network of the future. In such a network, signals remain in optical form while traversing fiber and convert to electrical impulses only at the network edge. This pure optical entity, say the visionaries, will achieve a measure of speed and flexibility that's impossible over the hybrid networks of today. It will enable application service models that are as yet undreamt of.
All-optical networks have been predicted since the early nineties, when the first wave division multiplexing (WDM) and optically based waveguides and filters began to appear. However, such networks remained conjectural, as there wasn't a means of switching an optical signal from one fiber to another without first converting the signal into electrical form.
MEMS is seen as the probable near-term market leader.
The Holy Grail now seems more tangible, as manufacturers have announced products ? such as symmetrical switches, cross-connects and add-drop multiplexers ? that are based on optical principles. A chief impediment to the realization of the pure optical network has apparently been removed.
"We see a rapid adoption of optical switch technology by the carriers," says Stephen Montgomery, president of the San Mateo-based ElectroniCast Corp., which has been tracking developments in fiber optics for more than two decades. "We think that the increases in traffic will dictate such a solution simply because electronic switches will not prove adequate or at least not cost effective. We also believe that proven technology exists to support such a change. Will that bring about an all optical network Not immediately, but it's an important step."
Such an assessment is, not surprisingly, enthusiastically endorsed by most of the manufacturers of optical switching equipment. And those manufacturers, given the near nonexistence of a market today, are surprisingly numerous. But faith in an optical switch solution is not shared by representatives of the more traditional electronic switching technologies.
"At this point it's all hype and marketing," snorts Nick DeVito, vice president of product management business at Tellium. "I'm not saying there aren't advantages to optical switching in principle, but right now, there are no real products."
DeVito cites a host of limitations in current optical switch technology. These limitations, admitted by at least some of the manufacturers themselves, are considered at length below.
Reasons for Caution
Tellium will compete with the vendors of photonic switches, and thus it might be expected to dismiss the new technology. Nonetheless, similar reservations about photonic switches have been voiced by individuals without an obvious stake in established switch technology.
"There is plenty of debate as to the real advantages of an all-optical network, as well as optical switches," states Pete Farmer, director of optical networks at Strategies Unlimited (Mountain View, CA), and author of a recent report on optical switch technologies. "People who make terabit routers tend to think that they'll be able to keep abreast of network demands, and frankly we don't predict the rapid abandonment of electronic switches. They are highly developed systems which will continue to get faster and better."
John Midgley, president of Lightwave Microsystems, a manufacturer of subassemblies for all-optical add-drop multiplexers, offers yet more reasons for caution in weighing the claims of the optical upstarts.
There are going to be some big losers in the switch business.
"I like to think I can be objective," he begins, "because we're in a related area, but aren't really competing with the switch manufacturers. All I can say is that it is really difficult to make predictions. What can be said is that there isn't a good optical switch on the market today. Which, if any, technology will become dominant If I had to bet, I'd say MEMS [micro-electrical-mechanical systems], but I'm not betting."
Jay Liebowitz, director of optical components at Ryan, Hankin and Kent, says, "even with extensive engineering knowledge in the area of fiber optics, it is very hard to make determinations concerning the effectiveness and competitiveness of the various approaches. The carriers have some inkling, but they're generally under nondisclosure. What we can say is that no technology has yet demonstrated Telcordia standards reliability, so we're not prepared to predict rapid adoption of optical switch technology at this time. Maybe in two years, you'll see fairly extensive deployments, but that's not certain."
Yet there is no disputing that optical switch manufacturers are numerous, attracting venture capital in substantial amounts. In some cases, these manufacturers are being acquired by major infrastructure companies for very large sums of money. Xros, for instance, was recently purchased by Nortel for over three billion dollars. No one denies that the current buzz about optical switches is deafening.
The What and the Why
The purpose of an optical switch is precisely the same as that of an electronic switch, namely, to direct traffic through the network. The difference between the two is that an electronic switch must convert light into electricity and then back again to effect switching or routing functions, whereas the optical switch maintains the signal in optical form.
Whereas electronic switches have always been based on common circuits, the optical varieties are extremely heterogeneous. More than a dozen techniques for switching optical signals have been devised, of which perhaps a half a dozen stand a reasonable chance of establishing a market presence.
"We see five technologies as currently viable," notes Leibowitz. "Those would be optomechanical, MEMS, liquid crystal, thermo-optical and Agilent's bubble technology."
The oldest and simplest optical switch type, and the only one to have found real employment in the field so far, is the optomechanical variety. In this type, fiber ends are mechanically aligned with one another by motors, much as in the old crossbar telephone switches used in predigital telephone exchanges. Optomechanical switches are currently manufactured by JDS-Uniphase, among others, and are used for restoration in some backbone networks, and also in cable television plants and in certain optical network test rigs.
Predictably, just about no one is assuming a place for the decidedly low-tech optomechanical switch solution at the core of the ultra-dense WDM networks of the future. Curiously, however, another mechanical system, MEMS, is widely seen as the probable near-term market leader, at least in the big central office switch market. Indeed, of all the individuals interviewed for this article, only David Anderson, director of R&D at Agilent's Optical Network Division, feels that MEMS can be successfully challenged in the large switch market.
"We can easily build 512 by 512 systems with our technology," Anderson insists. "Our platform is the best overall for large or small switches."
MEMS is a technology with a considerable pedigree. It is currently used in Texas Instruments' digital micromirror (DMM) video projection light engines, and in automotive air bag acceleration sensors, and has been under development for more than a decade. Apart from the optomechanical type of switch, the MEMS variety is the oldest.
As applied to optical networks, the MEMS concept is quite simple. Tiny mirrors mounted on pivots intercept the light signals conveyed via the individual strands of optical fiber coming into the switch. The mirrors are then tilted to direct the light beam into the desired fiber going out of the switch.
Two main types of MEMS switches exist today, the so-called digital or two-dimensional switch, and the so-called analog or multidimensional switch. None of these terms is entirely accurate, but they've passed into general industry parlance, so those terms are used here.
In a digital MEMS system, each mirror can assume only two positions and thus can only reflect light in two directions. Successions of mirrors must therefore be used to switch light across a fiber matrix. In an analog MEMS system, on the other hand, the mirrors can assume any number of tilts. Thus, a few mirrors working in tandem can reflect light into any of hundreds of individual fibers across the switch. The actual number of mirrors required in an analog switch is 2N the number of ports, as opposed to N squared in the case of the digital type. The precision of alignment required in an analog switch is orders of magnitude greater than for a digital switch, however.
Because size matters in a central office, analog designs are the only serious choices today for large switches. "Analog MEMS is where the action is," remarks Rajiv Ramaswami, author of a standard text on optical networks, and currently vice president of systems architecture at Xros.
Large analog MEMS switches have been announced by Xros, Lucent, Astarte, and Cronos. Optical Micro-Machines in San Diego has chosen to concentrate on small digital MEMS switches.
Present-day MEMS mirrors are constructed out of reflective silicon rather than metal. The devices are manufactured as large integrated circuits and thus presumably are economical. But one might question how reliable these devices will be over time. Such switches have complex physical structures with myriads of moving parts, all of which must remain in ultraprecise alignment.
"We're convinced that MEMS is a proven technology," asserts Stephen Montgomery of ElectroniCast
But others disagree. "It's definitely not proven," declares Leibowitz.
Steve Georgias, president of Network Photonics, an optical switch startup that has not publicly discussed its patented technology, notes that "the switches have to be made to submicron precision, and the mirrors have to occupy an individual position for long periods and then suddenly change states. What happens after a mirror has moved several million times Is it going to preserve alignment Remember, the light actually leaves the fiber and leaps a gap in these systems."
MEMS and optomechanical form one subset of optical switching technologies. They represent technologies that utilize mechanically actuated devices to manipulate light streams. All other optical switches use electro-optical or thermo-optical means to switch a signal.
The most prevalent methodology in this second group involves liquid crystal as the switching mechanism. Liquid crystal molecules, by definition, change their own polarization in the presence of a threshold voltage and either pass or block polarized light encountering them. While the initial polarization of an optical signal will not be maintained over distance, the 90-degree relationship between the opposing polarizations resulting from dispersion is constant. Each polarization may be separately aligned with that of the liquid crystal layer, a technique known as polarization diversity. In this manner, paths between opposing fiber ends can be opened and closed. As is the case with MEMS, a liquid crystal switch is essentially a succession of light valves.
Liquid crystal switches are currently manufactured by Chorum, Corning and Spectra Switch. Chorum and Spectra Switch claim to be close to shipping products.
Liquid crystal is taken seriously by most individuals in the industry, including nearly all of those in the MEMS camp. But almost no one, even liquid crystal advocates, sees the two technologies as directly competitive.
"Liquid crystal is not well suited to the construction of large optical cross-connect," says Scott Grout, president of Chorum. "Its real place is in optical restoration, add-drop multiplexers and cross-connects between two rings ? applications where one by two, two by two, and one by eight arrays are adequate."
Of course, small-scale MEMS switches are possible as well, but the liquid crystal camp believes that they hold the advantage. "We have no moving parts," observes Nick Lawrence, president of Spectra Switch.
The third technology in strong contention is exclusive to one company, Agilent, a subsidiary of Hewlett-Packard (HP). The technology is often inaccurately designated thermocapillary in reference to a related switch design developed by NTT; Agilent itself simply refers to its invention as a photonic switch.
The basic mechanism for the switch is borrowed from HP's well-known bubble jet printers. Switching takes place in a series of channels separating the fiber ends where bubbles exert diffractive effects on the laser light impinging upon them. The bubbles themselves are generated by heating the liquid in the channels. Exact details of how the angle of diffraction is controlled have not been disclosed, but the system has been demonstrated at trade shows and is reportedly being trialed now. Agilent has announced a third-quarter introduction.
Agilent's switch size is not comparable with the MEMS design ? only 32 x 32 ? but the company plans to compete in the large cross-connect market nonetheless. "You simply cascade the switches," explains Anderson.
"It won't work," objects Rajiv Ramaswami of Xros. "You suffer losses each time you cascade, and current optical amplifiers are wavelength specific so you'd have to have a lot of separate ones. In any event, cascaded systems don't scale linearly. The size of the switch and the cost of constructing it would be enormous."
Leibowitz expresses skepticism as well. "Making a really big switch could be a problem, but that may not be their market. We think there's a much bigger business in small switches. If the spec sheets are accurate, the Agilent system is very formidable. But we'll know more when the test results start coming in."
Many other optical switch technologies exist as well, involving such devices as optical waveguides and interferometers, semiconductor optical amplifiers (SOAs), lithium niobate, prisms, thermo-optical substances and so on. But unlike the three technologies considered above, none seem ready for the market.
Looking Forward
Whatever technology they espouse, all of the optical switch companies claim similar benefits. "You're no longer bit-rate limited," says Chorum's Grout, expressing the conventional wisdom, "and you avoid all of those optical to electrical to optical conversions, which is a big cost savings."
Many in the industry also believe that what is now a competitive advantage will soon constitute a decisive superiority.
"We're going to see pipes with a thousand and more fiber ports and hundreds of frequencies on each fiber," says Montgomery, arguing the imminent obsolescence of the big electronic switch. "We don't see how electronic switch technology can evolve to handle that."
Tellium's DeVito argues otherwise, however. "There's something important missing. These switches are basically just bundles of fat, dumb pipes. They can't read bits in the optical domain so they can't make decisions on what to do with traffic. And forget about all-optical packet routing."
"And they can't regenerate a signal, nor can they translate wavelengths," adds Grout playing the Devil's advocate for a moment.
When pressed, most manufacturers concede that optical switches alone won't bring about the all-optical millennium. Some question the need for such an entity.
"I think both electronic and optical network elements have their place," says Ramaswami. "Sure you want to eliminate conversions, but I don't see optical computing ever becoming a reality, which means you have to signal in the electronic domain. Use each technology where it's appropriate."
Which is precisely the issue at present. Electronic switching is the incumbent technology, while optical is the challenger and has yet to establish its appropriateness at any level.
"I don't think the optical switch industry is completely ready now," concedes Ramaswami, addressing that point, "but we're working hard to get there. One thing you can be sure of is that there are going to be some big losers in the switch business. It's just a question of who."
Dan Sweeney is an America's Network contributing editor.
americasnetwork.com |
