Startup Invents Laser Alternative

As the number of DWDM channels in optical networks multiplies, so do the number of fixed-wavelength lasers that are required to light them. This is a big problem for carriers, because it costs a lot of money to stock and manage inventory and spares for their systems. In fact, this problem prompted the development of widely tunable lasers (see Tune In!).

Early-stage startup KiloLambda Technologies Inc. has a new take on the problem. It is developing a laser subsystem that will deliver "hundreds or even thousands" of wavelengths simultaneously, according to its VP of marketing, Tzachi Ben-Gal. One of KiloLambda's subsystems could replace all the fixed-wavelength lasers in a DWDM system, he claims.

Too good to be true? Too early to say. Right now, KiloLambda's ideas are just that -- ideas. "We have 90 percent of the work done on the multiwavelength source, and expect to have prototypes in six months," says Ben-Gal.

It's worth pointing out that KiloLambda numbers only nine people, including the founders. The technology was invented by a father-and-son team, Moshe and Ram Oron, who started the company in February 2001.

But, despite its size, the company's founders appears to have influence -- Moshe Oron is the Israeli representative to the European Laser Network (Eulasnet), a collaborative effort among universities, R&D labs, and industry in Europe. He was previously the Israeli representative on the "Star Wars" project, President Reagan's misbegotten Strategic Defense Initiative of the 1980s.

Moshe, who holds the position of chief scientist, is also a part-time professor at the Technion-Israel Institute of Technology. His son, Ram, who recently gained a PhD from the Weizmann Institute of Science, is KiloLambda's CTO.

KiloLambda received an undisclosed amount of seed funding from Skypoint Capital Corp. and The Yozma Group. The startup says it has enough money to last two to three years and has no plans to grow until it makes its first sale.

Clearly, the company has its work cut out. Even if the product works as advertised, it still has to convince systems vendors and carriers of its usefulness.

KiloLambda has identified one application where its technology could give it the edge -- for increasing system capacity by using lots of channels crammed tightly together in frequency.

Right now, most carriers are considering OC768 (40 Gbit/s) systems as a way of increasing capacity while also cutting costs on their network. Although components for OC768 are more expensive than those for lower speeds, fewer components are required, so there should be an overall cost saving.

Other vendors have proposed using more channels rather than high-data rates to increase network capacity (see Essex Claims 4000-Channel DWDM and Essex Demos 40-Gig Alternative). But they always hit up against two problems. First, it's difficult to see how to reduce costs in high channel-count systems. When multiplying the number of channels, most of the components, and their associated costs, also have to be multiplied.

The second problem is that when channels are squashed together tightly, they start to interfere with each other. In order to get thousands of channels down a fiber, the data rate of each channel would have to be reduced to prevent crosstalk.

KiloLambda may have hit upon a way of solving, or at least mitigating, both difficulties. It's focusing on the lasers -- which makes sense, because lasers are one of the most expensive parts of an optical system. The startup is proposing a subsystem with just three basic building blocks. One of these is a laser; the company is keeping details of the other two under wraps.

As a result of volume advantages, its laser source might prove to be more cost effective to manufacture than today's fixed-wavelength lasers. In addition, only one laser source is needed per subsystem, and the two mystery building blocks might work out a lot cheaper to make than lasers -- although, of course, it's too early to tell if this will actually turn out to be the case.

KiloLambda claims that another key feature of its subsystem is that all the channels are extremely stable with respect to one another. "You only need to stabilize one channel, and the whole source is stabilized," says Ben-Gal. That's important, he explains, because it allows system bandwidth to be used more efficiently -- there's no need to leave dead space or "guard bands" between channels.

In the initial phase of development, KiloLambda plans to use channel spacings of 10 GHz to carry 10-Gbit/s signals -- for 100 percent bandwidth utilization. The International Telecommunication Union (ITU) standard for 100GHz channel spacing looks pretty inefficient by comparison. Every wavelength is allowed to move independently by 20 percent with respect to its center value, which effectively means that 40 percent of bandwidth is wasted.

— Pauline Rigby, Senior Editor, Light Reading
grateful photon 12/4/2012 | 10:38:01 PM
re: Startup Invents Laser Alternative
agreed, an ultrafast laser pulse does this. also, a supercontinuum laser could provide broad, coherent radiation across the C band and more. regarding the 100% utilization, it's fundamentally limited by shannon's theorem, and practically limited by non-ideal behavior of devices.

in both cases, VOA arrays or the like for flattening would be necessary, and don't forget that each lambda must be modulated independently. considering the low quantum efficiency across such a large chuck of spectrum, you could imagine some gain would also be required in the form of EDFAs or SOAs. by the time you put it all together, the cost savings may evaporate. remember that discrete lasers are not all bad, allowing for channel replacement etc. i would rather replace a spark plug than have to replace the entire electronic ignition module on my car, if one cylinder isn't firing.

such systems were not invented by these guys. modelocked, femto fiber lasers have been available from IMRE for years. unless they have a breakthrough on cost of implementation, other folks have taken a pass on this approach due to cost and complexity offsetting performance gains.

it feels like another optical startup 'selling' known technology to VC's who think it's new and enabling. walking around OFC, i saw plenty of examples of that from the last few years. why should it stop now, have VCs become optically savvy? don't answer that, i think we all know the answer.

arch_dude 12/4/2012 | 10:38:01 PM
re: Startup Invents Laser Alternative google for femtosecond comb. This technology does
exactly what this company is talking about.

The main problem: the non-linear optical
fiber it depends on can only accept a modest
power density. Then it melts. If they solve that,
then you really do get the "%100 percent"

Of course the "100%" is jus a ratio of
bps/hz. There is no law that restricts you
to 1bps/hz. In the electrical domain, there is
a law that restricts you to 1 baud/hz, but you can
have multiple bits/baud. It's not clear to me that
photons work exactly the same way.
let-there-be-light 12/4/2012 | 10:38:00 PM
re: Startup Invents Laser Alternative Did I miss something? How would you modulate the indvidual channels? Or is the answer obvious? It isn't to me. anyway.
dwdm2 12/4/2012 | 10:37:59 PM
re: Startup Invents Laser Alternative "In the initial phase of development, KiloLambda plans to use channel spacings of 10 GHz to carry 10-Gbit/s signals -- for 100 percent bandwidth utilization. ..."

KiloLambda's approach, on the face of it, looks attractive.

However, a 10 Gb/s signal on a 10 GHz grid gives a spectral efficiency of 1 bit/s/Hz. On a different thread we have discussed this issue. The highest spectral efficiency reported so far is 1.24 bit/s/Hz.

Would someone please explain how this can translate to 100% bandwidth utilization?

dwdm2 12/4/2012 | 10:37:58 PM
re: Startup Invents Laser Alternative It is hard for a rational one to conceive that the VCs would be so gullible... . Although over the last few years the optical communication sector has shown that the circumstance is pretty much as described by grateful photon (see below). When it comes to invest for profit, I can't believe that VCs are doing their due dilligence. Perhaps there is some other explanation...

Just a thought

"it feels like another optical startup 'selling' known technology to VC's who think it's new and enabling. walking around OFC, i saw plenty of examples of that from the last few years. why should it stop now, have VCs become optically savvy? don't answer that, i think we all know the answer."
grateful photon 12/4/2012 | 10:37:52 PM
re: Startup Invents Laser Alternative my guess...

brute force approach would be to use discrete modulators (EA or EO, for example) on each channel. this means at some point the lambdas have to be split out with a demux or at least free-space coupled to fibers from the dispersive element from which they originate. not cheap in parts or assembly.

if one took the femto pulse and dispersed it with a grating, a very fast SLM (don't know where to get one for 10Gig) could work with free-space coupling and another grating to mux back together.

of course, this group may just want to deliver a CW source, leaving the modulation and power management to the systems guys. again, big issue compared to discretes is service and rel. when your broadband laser craps out (infamous behavior for finicky femtos) you lose all lambdas for this one failure -- ouch!

let-there-be-light 12/4/2012 | 10:37:50 PM
re: Startup Invents Laser Alternative Sounds a bit like expecting an octopus be able to swim in at least 8 directions at the same time, simply because it has 8 "arms" (what are those protuberances called officially?)
catbrier 12/4/2012 | 10:37:37 PM
re: Startup Invents Laser Alternative grateful photon: "brute force approach would be to use discrete modulators (EA or EO, for example) on each channel..."

Wayne Knox & coworkers at Bell Labs did something like this a few years back to develop a low-cost fiber-to-the-home system. They used a mode-locked erbium fiber laser to generate short pulses (broad sprectrum), then chirped it to spread it out in time. I believe they used a highly dispersive fiber, but a grating pair would probably also work. Then they ran the stretched pulse through a modulator. The electrical input to the modulator was a TDM stream but since the pulse has been chirped, this results in WDM modulation. One then de-chirps the pulse to shorten it and puts it into the transmission fiber.

Possibly the great new breakthrough announced in this article is something along these line. Mode locking or supercontinuum generation are common ways of generating broad spectrum pulses.

The big downside is the low duty factor (pulse rep rate of the mode-locked laser). I think Knox & compay addressed this by splitting the original pulse into multiple arms, modulating each separately, time delaying each arm, and then interleaving them after de-chirping. It's better than having a modulator for each wavelength, but still gets very complicated to achieve a reasonable duty factor.

A more accurate account is available in their paper:
A 1021 channel WDM system,
Collings, B.C.; Mitchell, M.L.; Boivin, L.; Knox, W.H., IEEE Photonics Technology Letters, Volume: 12 Issue: 7 , July 2000, Page(s): 906 -908.
billyoung 12/4/2012 | 10:37:32 PM
re: Startup Invents Laser Alternative 1000 DWDM channel @ 10Gb/s is 10Tb/s, can one modulator do it?

BTW: 10Gb/s occupies at least around 20GHz optical spectrum. special modulation format and filters are needed if really wanted to be in 10GHz window -> more cost, not less + DWDM issues
Sign In