x
Optical/IP

GigaTera Goes Multiwavelength

Swiss startup GigaTera AG is pursuing a new application for its pulse-generating laser technology -- one that could make it possible to cram 32 channels onto a single line card (see GigaTera Demos Multi-Wavelength Laser).

GigaTera unveiled its pulse-generating laser, called ERGO, last year. In a nutshell, the device produces a train of RZ (return-to-zero) pulses at 40 Gbit/s without requiring a separate modulator (see GigaTera Sends Pulses Racing). It turns out that the same technology, with only a minor modification, can be used to make a device that delivers 32 channels of light simultaneously.

"All we have to do is alter the pulse repetition rate of the laser," explains Michael Brownell, the company's VP of product development. "The rest is automatic."

It's not immediately obvious how a device that can produce pulses can also produce multiple continuous-wave (not pulsed) channels of light. The answer lies in a mathematical tool-cum-physical phenomenon called "Fourier transform." Understanding this is not easy or intuitive, but here's a quick explanation for the physicists among you.

Ultra-fast pulses of light can be split up -- both theoretically and literally -- into a number of frequency components. When the pulses have a fixed repetition rate, then those components consist of a series of individual frequencies with the same spacing as the repetition rate. Therefore, a laser that pulses at 25GHz will give rise to channels spaced by 25GHz, which happens to match the International Telecommunication Union (ITU) grid.

Whether the user sees pulses or multiple channels depends on how the light is measured, says Brownell. If the laser beam is observed as is, without splitting or changing the output in any way, then it consists of pulses. If the different frequencies are separated, by sending the laser beam through a demultiplexer, then each output of the demultiplexer yields a separate channel.

GigaTera doesn't claim to be first with this idea. In fact, researchers at Bell Labs demonstrated something similar six years ago, although it doesn't appear to have commercialized the work (see Bell Labs' press release for details).

"We can't claim credit for inventing this," says Brownell. "But the key here is that our device is super-simple, stable, small, and cheap."

The device can be small and cheap because it contains considerably fewer parts than 32 fixed-wavelength light sources. An equivalent fixed-wavelength solution would require 32 lasers, 32 sets of driver electronics, and 32 wavelength lockers. A multiwavelength light source comprises just four components: a single pulse-generating laser, one set of driver electronics, one wavelength locker, plus something extra -- namely a dynamic gain equalizer (DGE), which demultiplexes the channels and then balances the output power to ensure it is the same across all channels. GigaTera is using a DGE from Silicon Light Machines for this purpose.

The ERGO laser is currently available as a chassis-based instrument, which potential customers can use to evaluate the technology. GigaTera says it can bring the multiwavelength laser source to market quickly by building it into the same chassis.

In the future, the company plans to shrink the instrument down to something that will fit onto a line card. This should be feasible, given that the laser alone comes in a tiny package measuring approximately 15 mm per side.

Analysts think it's an intriguing idea, but many wonder if it will catch on. "Yes, it is very interesting," says Tom Hausken, director of optical communications components research at Strategies Unlimited. "However, I think the fact that Bell Labs didn't pursue it says a lot."

There are a number of reasons for potential customers to be wary of the new technology, he says. The main one is that they would be putting all their eggs in one basket: If the device fails, a carrier loses 32 channels in one fell swoop, instead of just one.

The idea of putting 32 channels on a single line card may also be a bit ahead of its time, in the sense that there's a long way to go before systems vendors can cram the electronics associated with 32 channels onto a single line card. Most systems today still have only a single channel per line card.

Nevertheless, GigaTera's Brownell insists that people are interested in the multiwavelength light source. "We've pushed ahead with its development as a result of customer feedback," he says. One of the early applications is likely to be in test and measurement, which wouldn't require the device to be shrunk down onto a line card.

It's worth pointing out that at least one other startup -- Israel's KiloLambda Technologies Inc. -- is developing a multiwavelength light source (see Startup Invents Laser Alternative). So far, however, KiloLambda hasn't revealed much about its technology.

— Pauline Rigby, Senior Editor, Light Reading
Page 1 / 2   >   >>
kspinka 12/5/2012 | 12:51:41 AM
re: GigaTera Goes Multiwavelength In the article there seems to be an invalid concern that if the device fails you will lose all 32 channels. Of course you will, but that is analogous to saying that it is a risk to scale signals from 1.5Mbps to 10 or 40Gbps. If you lose a 10 or 40Gbps signal, you are potentially losing many more "phone calls", or data packets than if you lose a 1.5Mbps signal. However, with all new density paradigms, there can always be an efficient, redundant element designed into the actual deployment.
If a productized version of this technology was to fit in say 2 RU (3.5 vertical inches), this would become a power, cooling, real estate, and of course laser transmitter, cost benefit.
I am of course very curious as to the optical power output and MTBF for this laser as well as any long-duration operating issues, such as drifting, etc.
In my opinion, this is a fantastic development. It is further reducing the cost-per-bit-transmitted from a different product-component dimension.

/kristofer
holden_caufield 12/5/2012 | 12:51:40 AM
re: GigaTera Goes Multiwavelength Do a search on IEEE Explore for keywords such as SuperContinuum Broadband and you can see how long the Japanese have been experimenting with a single multi-channel source, including ULH transmission results.

(Whoosh! More VC money flushed.)
isitso? 12/5/2012 | 12:51:39 AM
re: GigaTera Goes Multiwavelength This concept is in fact very very old. The savings are questionable since you need to amplify your power right after the wavelength separation. For that you need an EDFA. So, why use a lousy mode-locked semiconductor laser + EDFA rather than a nice mode-locked Er laser to begin with. This is EXACTLY what the Japanese and LUcent and NRL and everyone else had been doing for ages.
The efficiency of the scheme is low since the output spectrum is Gaussian, so when you try to
cut 32 channels of roughly equal intensity you
loose the "tails" of distribution. The power equalization is a big problem. The RIN of each spectral line is horrendous sinc ethe power is so low. Most important you still need 32 external modulators and that is where the real cost is.
But hey, if VC's are ready to through money on cool research projects, God bless them! Maybe some suckers will waste money on some of the stuff I am doing in my Univesity lab!
Actually, that being Europe, I can see a strong Government hand behind this ludicrous thing.
redface 12/5/2012 | 12:51:35 AM
re: GigaTera Goes Multiwavelength I don't think you just need one set of driver electronics to transmit all 32 channels of information. The pulsed laser + demultiplexer give you 32 streams of continuous light, but you need to modulate each light beam individually to transmit signal and therefore 32 sets of driver electronics are needed. So the savings is really only in the laser transmitters and it is much smaller than stated in this article.
photonsu 12/5/2012 | 12:51:34 AM
re: GigaTera Goes Multiwavelength Hey "isitso, have YOU seen it? I doubt it, because I have. Don't speak if you don't know what your talking about. You add nothing, and It'll save bandwidth.
photonsu 12/5/2012 | 12:51:33 AM
re: GigaTera Goes Multiwavelength 1) One device replaces 32.
2) Two are needed for redundancy rather than 64 lasers. (Yes, tunables reduce sparing count but needs optical switch)
3) Wavelengths are harmonically locked eliminating wavelength lockers totally.
4) Spectrum can be amplified and split multiple times so a single source might serve a CO.
5) Power, space, heat and $$ savings.
6) Enhanced reliability.

As noted, each channel now needs a modulator, but all modulators can be the same, wavelength independent, rate and format independent, and produce enhanced spectral characteristic leading to greater range potential or elimination of (some) dispersion compensation. This will allow economy of scale manufacturing of modulators and reduced DWDM implementation cost.
isitso? 12/5/2012 | 12:51:33 AM
re: GigaTera Goes Multiwavelength "As noted, each channel now needs a modulator, but all modulators can be the same, wavelength independent, rate and format independent, and produce enhanced spectral characteristic leading to greater range potential or elimination of (some) dispersion compensation. This will allow economy of scale manufacturing of modulators and reduced DWDM implementation cost."

And what the hell are you talking about, man?
In every system operating now all teh modulators are exactly the same , wavelength-independent,
so where are the savings???
isitso? 12/5/2012 | 12:51:33 AM
re: GigaTera Goes Multiwavelength I did
photonsu 12/5/2012 | 12:51:32 AM
re: GigaTera Goes Multiwavelength Yes, you are correct. But ...
The savings comes from the fact that EA modulators can now produced in arrays and with higher yield since the complexity of the process required to make EML like devices is simplified.
Vesting 12/5/2012 | 12:51:27 AM
re: GigaTera Goes Multiwavelength Even if they can transmit up to 40 wavelengths using a single laser I don't see the commercial application in the current economic environment. Most DWDM systems deployed today are operating well below their defined capacities.

A laser that can support 40 wavelengths that is only being utilized to provide less than 10 really hasn't bought much.

Until the bandwidth glut goes away demand for any type of wdm is going to be week.

Page 1 / 2   >   >>
HOME
Sign In
SEARCH
CLOSE
MORE
CLOSE