Network Photonics Lifts the Lid
Today, the startup is out to prove that it hasn’t thrown the baby out with the bath water. It’s lifted the lid on the “CrossWave” technology behind its switch for the first time and it’s explained why it thinks this gives it a big edge over other developers of all-optical wavelength switches (see Network Photonics Tells All).
In essence, it comes down to having a switch that's much smaller and less power hungry than competing developments, and yet can scale to handle many thousands of wavelengths.
Network Photonics also says that its switch chips were designed from the ground up to be easy to manufacture. As a result, the company has already achieved yields of “better than 50 percent” on the “thousands” of chips it’s already made, according to Steve Georgis, the startup’s chairman, president and CEO. “We expect it to be better than 90 percent when we’re in volume,” he adds.
The bottom line? The cost of Network Photonics’ switches will be “something less than one tenth” of developments based on other technologies, according to Georgis.
The other big message that Georgis wants to get over is that Network Photonics hasn’t lost much by abandoning its original concept of developing a complete package of technologies including not just its all-optical switch but also a transmission system and management software.
Georgis says that his switch can handle all of the 80 wavelengths in the “C” band of the grid defined by the International Telecommunication Union (ITU). The fact that these are standardized means that Network Photonics’ switch will interoperate successfully with other vendors’ transmission systems, he contends.
“The carriers were saying we really just want the switch,” says Georgis. They want to carry on using DWDM systems from their established suppliers, but add Network Photonics’ switch so they reconfigure their networks on the fly. “Most of the big carriers are working with us,” Georgis adds.
Network Photonics has already conducted interoperability tests with other vendors and is discussing partnerships with some of them, according to Georgis. Calient Networks Inc., which is developing a "transparent" high capacity optical switch that will handle any spacing of wavelengths in any waveband. "We really want to make something that won't become obsolete when the next DWDM systems come out," he says.
So, what’s inside Network Photonics’ switch?
It’s tiny tilting mirrors based on MEMS (micro-electromechanical system) technology. But it isn’t the 3D MEMS championed by Calient and others (see MEMX Starts Anew on 3D MEMS). It’s not even 2D MEMS, the technology used in smaller scale optical switching subsystems made by the likes of OMM Inc. It’s what Network Photonics calls “1D MEMS”.
Network Photonics gets its cute “1D” marketing moniker from the fact that its switching fabric is just a single line of up 80 tiny tilting mirrors (a 1 by 80 array), constructed from gold plated silicon on a single, small, chip. This is combined with a gold-plated silica “dispersive element” which acts like a prism, splitting incoming light from the same fiber into the 80 wavelengths in the ITU grid and steering each one onto its own particular mirror. It also does the reverse - recombining reflected wavelengths and carrying them into the appropriate output fiber.
As the mirrors simply flap up and down, controlling them can be done digitally, as is done with 2D MEMS. It avoids the complexities of trying to develop analog controls for the two axis of rotation of 3D MEMS mirrors – one of the main challenges facing developers of this type of switching fabric. “Everybody is having a hell of a time trying to control these things,” says Georgis.
Calient's Bowers acknowledges that developing control systems for 3D MEMS switches is challenging but says "it's do-able if you have the right technology."
Network Photonics says that its design means that only one mirror is needed for every wavelength handled by the switch. That compares with two mirrors per wavelength for 3D MEMS switches and the number of wavelengths squared for 2D MEMS switches. In other words, Network Photonics can make a bigger capacity switch with less parts.
“We’ve got designs that take us up to several thousand by several thousand wavelengths,” says Georgis. “We have the simplicity of 2D – digital control – with the scalability of 3D,” he adds.
Calient's Bowers points out that the capacity of switches is usually measured in terms of the number of states they support. Calient's switch can handle 64,000 switching states, he adds. In other words, up to 64,000 connections can be handled without any risk of one connection blocking another. Network Photonics' chips appear to support 80 states each, so it would need 800 to match Calient's non-blocking capacity.
Bowers also notes that in Calient's case, each connection can be a single wavelength or a group of wavelengths, up to all of the wavelengths from one fiber. Network Photonics' switch, on the other hand, only handles single wavelengths - another issue to consider when reviewing overall capacity.
It’s not just the reduced number of mirrors that helps Network Photonics achieve its density claims. Its “dispersive element” inside the box means there’s no need for separate equipment for demultiplexing and remultiplexing wavelengths either side of the switching core, as is often the case with other designs. The upshot is that Network Photonics’ switch occupies five times less space than switches using alternative technologies. It also uses far less power according to Georgis.
Right now, the company is developing an 80 by 80 wavelength switch. This can be used on its own, or several of them can be linked together to make a bigger capacity switch.
One of the snags could be that each 80 by 80 switch has a loss of 5 decibels, and this will accumulate in bigger switches comprising several boxes. Georgis says some 2D and 3D MEMS switches have even higher losses – something that is acknowledged by Conrad Burke, OMM's senior VP of marketing and business development.
Georgis also says losses aren’t much of an issue in many cases because Network Photonics’ switch typically will be installed at an amplifier site, so boosting the light power after the switch isn’t a big deal.
Network Photonics’ claims look good on paper, but of course, the proof of the pudding will be in the eating. It’s already made evaluation units and expects to start beta trials in December. General availability is scheduled for March of next year.
By then, some of the drawbacks might become clearer and some other startups might have sprung some surprises. Watch out, for instance, for UK startup Polatis Ltd., which is also planning to lift the lid on its technology in the coming weeks.
Then again, a lot of startups in this field are likely to go belly up. Georgis reckons that he’s done what he needs to ensure Network Photonics’ survival. “We need to ride out the storm,” he says, noting that Network Photonics has now raised $139 million in funding, some of it only recently (see Network Photonics Scores $20M). “We’re one of the best financed startups in the industry right now,” he says. — Peter Heywood, Founding Editor, Light Reading