OEOs are AOK, Startup Says

“Everybody has known that integration was going to make sense for photonics the same way it made sense for electronics 45 years ago,” says Rick Dodd, product marketing manager for Infinera, in an August interview with xchange.

“Now, that technology actually exists, it will be in carrier networks this year. In my mind, it is a very big step in terms of the evolution of how WDM has done.”

Optical-electrical-optical conversions, or OEOs, as they are called, earned their bad rep for being expensive. Scott Clavenna, chief analyst for research firm Heavy Reading, explains: “The driving force of all those optical innovations from ’98 until now was based on the assumption that when you stopped and did an electronic regeneration, you were spending a lot more money per bit than you would if you could push the photons along in an optical domain.”

This produced better amplifiers, better detectors, better gain management and other technologies designed to make the signal as clear as possible for the longest distance possible.

The drawback with the all-optical approach is that in eliminating the electrical, it also eliminates the digital, which is where much of the network functionality is introduced. Of course, the industry has been trying to replicate this in an all-optical manner, hence, the heralded debut of the ROADM, which does manage some of these functions, but falls short of wavelength conversion and multicasting made possible in the electrical world.

“OEOs are wonderful from a functionality point of view because that is the only way you go from optics to electronics. In electronics we can do digital processing,” says Dodd. “The beef that the industry has had with OEO costs is spot on. We can solve the cost problem without getting rid of the OEOs. The way we do that is we make OEOs very low cost.”

OK, that sounds simple enough. Surely, others thought of it before. They have and did, but it was mostly dismissed as not being doable with technology available at the time.

One vendor, Meriton Networks Inc., addresses the problem in its 7200 OADX by executing OEO regeneration without breaking the SONET signal down to its lowest granularity. “When we do an OEO and a regen on the signal, we do it serially; we don’t look at the protocol. We look at the bit stream and we clean up the signal,” says Coleman Hum, product manager for Meriton, explaining that its process includes reshaping, retiming and reamplifying the digital wave form. “By doing those three, we don’t have to look at the protocol. When the signal comes out it is as good as the original and at a very low cost.”

Analyst Clavenna says Meriton’s approach, while effective for metro applications, does not pull out any of the management overhead. “If you don’t do that, it does limit the operator’s ability to operate their network with the same level of monitoring as they would have in a SONET network today,” he says.

Another approach, the one taken by Infinera and the one that has frustrated the industry for years, is photonic integration. “Photonic integration basically means you can take lasers, modulators and other devices and put them all together on one chip, but also you can have many wavelengths of those functions on one chip as well,” explains Dodd.

There are a plethora of other optical integrated circuits that collapse one or two functions, but none attempts electrical regeneration. This is because indium phosphide, the material used to make the integrated circuits, is generally considered difficult to work with for electronics and optics, says Clavenna. “It’s very brittle; it’s very difficult to get uniform performance across the entire chip. It’s difficult to tightly pack all those functions onto a single piece because some functions like the properties of different materials. Most people thought if you tried to do everything on one piece of indium phosphide some of the functions would be compromised.”

Infinera, apparently, has proven them wrong.The company came out of stealth mode in May by launching the first digital optical networking system, Infinera DTN, based on large-scale photonic integrated circuits that it developed in house.

Infinera’s photonic integrated circuits include a 100gbps transmitter and 100gbps receiver. The transmitter integrates 10 lasers, 10 10gbps modulators and an optical multiplexer. The receiver integrates an optical demultiplexer and 10 photodiodes. Each enables optical-electrical conversion on a semiconductor chip no larger than 5mm on a side.

Infinera DTN offers digital control of optical networking functions such as adding and dropping customer traffic, managing bandwidth, monitoring performance, and eliminating noise and optical impairments.

The system comes in at about half the cost of existing systems. “If you can come in at half or even a third of the price, then you really do upset the balance, I would say,” says Clavenna. “That would change the competitive landscape.”

But it’s not a sure thing. Service providers, OEMs and chipmakers are skeptical about whether Infinera can do what it says.

Dodd says the company’s systems currently are in carrier lab trials and are expected to begin carrying live traffic by the end of the year.

“If they were to land a Sprint or WilTel, then that really puts them up there on the radar. That would make these other vendors at least have to propose an answer,” says Clavenna, referring to the large OEMs like Nortel and Lucent as well as other chipmakers like JDS Uniphase or Bookham.

Share this article: Email, Slashdot, Digg, Del.icio.us, Yahoo!MyWeb, Windows Live Favorites, Furl
Add this article feed to: RSS, My Yahoo, Newsgator, Bloglines