& cplSiteName &

Out of the Lab: Storing Light

Light Reading
News Analysis
Light Reading
1/22/2001

It’s often said that genuine optical routers won’t arrive until there’s been a fundamental breakthrough in physics -- until someone has figured out a way of buffering light so that packets can be stored and forwarded the way they are in today’s electronic routers.

Well, it looks as though we've got that fundamental breakthrough. In fact, two independent groups of scientists claim to have managed to slow down and "stop" light -- and both groups are aiming on publishing their findings this month.

The first paper, due to be published on Thursday, Jan. 25, in the journal Nature, describes how scientists from the Rowland Institute of Science in Cambridge, Mass., have arrested a pulse of light inside a gas of cold sodium atoms.

The second paper, due to be published next Monday, Jan. 29, in the journal Physical Review Letters (PRL) gives results of a similar experiment conducted by scientists at the Harvard-Smithsonian Center for Astrophysics.

The fact that two independent outfits have come up with similar results is significant. While the process of peer review in scientific journals weeds out most of the crackpots, there have been a few instances -- like cold fusion -- where work has proved impossible to reproduce, and thus of no practical use. This appears unlikely to happen with this research.

The key difference between the two experiments was the material used. Scientists at the Rowland's Institute employed a gas of cold sodium atoms, while the researchers at the Harvard-Smithsonian worked with a glass cell containing rubidium, which they heat up to create rubidium vapour.

In both cases, the atoms in the gas normally absorb light -- in other words, the gas is opaque. It can be made transparent (non-absorbing) to a particular wavelength by illuminating it with a so-called coupling laser. If the coupling laser gets turned off as a pulse is passing through the gas cloud, the pulse "stops." Turn the coupling laser back on, and the pulse continues on its journey. If information is encoded on the pulse, it can be recovered later, just like a letter delayed in the mail.

The idea of stopping light is one that would curl Einstein's hair. The trick is that the light in the pulse isn’t actually stopped at all. Instead, the information in the pulse has been transferred to the surrounding gas atoms, while the energy it contains passes into the coupling beam. The information stays trapped in the atoms until they get a kick of energy from the coupling laser being turned back on. Then -- kapow! -- the pulse springs back to life.

Of course, there's a lot of work to be done before it will be possible to engineer this technology into something practical. "This technology could be 10 years from application, it could be 50," says David Phillips, lead author on the PRL paper.

He adds, "There are still a bunch of physics issues to understand before you start engineering it into a box." Experiments that take place at very low temperatures -- like the work at Harvard-Smithsonian -- are useful in this respect because they are a lot easier to understand from a physics perspective."

Pretty soon, it should be possible to store light in solid materials, as well as in gases. "For real applications, solid state is where you'll think about doing things," notes Phillips. In fact, he believes that researchers are already submitting proposals to do these experiments -- storing light -- inside doped optical fiber.

— Pauline Rigby, senior editor, Light Reading http://www.lightreading.com

(11)  | 
Comment  | 
Print  | 
Newest First  |  Oldest First  |  Threaded View        ADD A COMMENT
Page 1 / 2   >   >>
Pauline Rigby
Pauline Rigby
12/4/2012 | 8:59:47 PM
re: Out of the Lab: Storing Light
Does anyone have an opinion on whether it will be nearer 10 or nearer 50 years before this technology gets to market?
Peter Heywood
Peter Heywood
12/4/2012 | 8:59:47 PM
re: Out of the Lab: Storing Light
Sounds like it to me, but perhaps there's a big catch. Is there?
FarAway
FarAway
12/4/2012 | 8:59:46 PM
re: Out of the Lab: Storing Light
Even if it took 10 yrs to make it practical to store some bits off light, how many more yrs does it take to make an all optical processor that can read, write and calculate those bits stored in light? Light is fast to transport, is it also fast to calculate? It might be a long long time before we can build all optical routers.
Peter Heywood
Peter Heywood
12/4/2012 | 8:59:43 PM
re: Out of the Lab: Storing Light
Isn't this argument about not being able to calculate things optically a bit of a red herring?

Wouldn't it be possible to make a router that genuinely preserved packets as pulses of light as they passed from the input to the output port and yet did all the processing to figure out where to route the packet electronically?

I guess it all depends why you're trying to make an optical router. If it's to eliminate expensive OEO interfaces, then this argument holds up, doesn't it? Maybe it doesn't hold up if you're trying to make routers that work very fast.



Katie
Katie
12/4/2012 | 8:59:42 PM
re: Out of the Lab: Storing Light
Surely the whole idea of replacing OEO with all-optical is to eliminate the speed bottleneck caused by the electrical portion of the router?
If these routers are to be used in the backbone network, then cost isn't such an issue, since a relatively small number will be needed. It's when we move further into the access portion of the network that money becomes an issue, since more of these devices will be needed.

I reckon it's possible to design a router that can use the information encoded in the light pulses to route the packet, without having to convert to electrical. Innovative companies like Bookham Technology and Ilotron should be watched closely, since they are likely to be leading the way in developing this type of technology.
magnani
magnani
12/4/2012 | 8:59:40 PM
re: Out of the Lab: Storing Light
So the argument about whether it's necessary to do the actual computation optically or just keep the data path itself optical can be a bit of a red herring or it could be important. It's significantly easier to do just the data path optically and control it electronically if we're talking about TDM (time division multiplexed) traffic where the connections are pretty much nailed down... but we are moving towards packet-switched networks where the data contains the information for where it's going... if you're going to do a pure optical setup on a packet switched network then it would be nice to be able to do optical computation.

People have been working on this for decades, and for some kinds of things it's truly stunning what can be done optically... but the advantage is not that the stuff moves at the speed of light... sure, the speed of light in fiber is a little faster than the speed of light in copper (I think it's something like 0.85c vs. about 0.65c or something close to that...) So if someone wants to do the computing optically for speed, it's a weak argument... if the routing information is in the data, then it would be nice to not have to waste money on transceivers to get that information.

The real advantage to optical computing is when you do it in an entirely different way... For instance, there a very common, useful computation called an FFT (Fast Fourier Transform) The time to do this calculation increases with the size of the data set you are working on. People have demonstrated a way to convert the data into a light pattern (no matter how big or small) and just shine that pattern into some special optics, and the answer shines out on the other side in constant time no matter how big the data set is... that's impressive... but if you want to use optical components to build computation engines like we have traditionally, that's a lot harder, and it's not clear it can buy you much other than saving the conversion time and cost...

Fundamentally to do traditional computers/microprocessors/logic, you need a technology that has an inherant, natural non-linearity... an inverter... show a component that blocks light when light is shined on it, and transmits light when there is not light hitting it... and I'll show you how to do the rest... Now make that part about the size and power of conventional integrated circuits, say something a little smaller than half a micron (a millionth of a millimeter) and now you've got competition for the conventional integrated circuit. I personally think the optics stuff is cool, but we're going to be living with IC's an awfully long time... no need to continue selling your Intel in a furry for optical startups... there's still a lot of work to make it reasonable.

So a bit long winded, but to get to the original question... will we see the holy grail of a pure optical network soon and how does this affect that? Well, if you just care about steering the traffic, then you don't even need this technology, and we might see this in about 10 years. If you want to make a true pure optical system and you want to do it with traditional von Neuman computers, then this technology is essential, but there's a whole lot of other missing pieces, so we MIGHT see that in fifty years, or it might not ever make sense... It needs a breakthrough, either in the method of doing the computational work, or the technology to carry out the traditional solution... Hard to say. A truly pure, optical system may just never make sense.

Just my opinion of course...
Katie
Katie
12/4/2012 | 8:59:37 PM
re: Out of the Lab: Storing Light
I was interested to read the previous posting, and agree with much of it. However, the bottleneck in the OEO section is not simply a case of electrons through silicon being slower than light down a fiber, it's also to do with the number of channels that arrive at the router.

If there are 100 channels arriving at the router, then the electronic circuitry is set up to handle 100 channels at an optimum speed. However, if we use techniques such as DWDM to upgrade the capacity of the optical network and push down, say, ten times more channels, then the electronic circuitry is unlikely to be able to cope with the extra traffic without a significant slowing down of throughput.

Of course, there's no reason why the network operator couldn't put in more channel processing capability than is currently needed, but in these days of cutting costs to the bone, it's unlikely that they will want to do too much over-provisioning.

However, the beauty of all-optical systems is that, as long as the optical router is correctly designed and set up, it should be able to handle as many channels as are transmitted (within a certain pre-set optical bandwidth of course, and with the necessary compensations for wavelength-dependent attenuation effects) - altogether a much better proposition for the network operator.

I'm not saying that all-optical routers are perfect - I'm sure they too have their drawbacks. But in the long-run, the benefits of having an end-to-end all-optical network are likely to be greater than having to include electronic components.
FarAway
FarAway
12/4/2012 | 8:59:24 PM
re: Out of the Lab: Storing Light
Eliminating expensive OEO is one thing, and coming up with a pure optical router is another. Eliminating OEO does not require light buffering.

FarAway
FarAway
12/4/2012 | 8:59:23 PM
re: Out of the Lab: Storing Light
>I reckon it's possible to design a router that
>can use the information encoded in the light
>pulses to route the packet, without having to
>convert to electrical.

I guess it is how LONG does it take for people to come up with this kind of technology that we were talking about.

Remember it's said in the article that those scientists used some kind of steam to capture sort of an image of the light. The light in fact passed through. It seems apparent one has to look at the information captured and do some calculation before deciding where to route the captured "packet".




FarAway
FarAway
12/4/2012 | 8:59:23 PM
re: Out of the Lab: Storing Light
>I reckon it's possible to design a router that >can use the information encoded in the light >pulses to route the packet, without having to >convert to electrical.

I guess it is how LONG does it take for people to come up with this kind of technology that we were talking about.

Remember it's said in the article that those scientists used some kind of steam to capture sort of an image of the light. The light in fact passed through. It seems apparent one has to look at the information captured and do some calculation before deciding where to route the captured "packet".




Page 1 / 2   >   >>
Featured Video
Upcoming Live Events
October 22, 2019, Los Angeles, CA
November 5, 2019, London, England
November 7, 2019, London, UK
November 14, 2019, Maritim Hotel, Berlin
December 3-5, 2019, Vienna, Austria
December 3, 2019, New York, New York
March 16-18, 2020, Embassy Suites, Denver, Colorado
May 18-20, 2020, Irving Convention Center, Dallas, TX
All Upcoming Live Events