'Corkscrew' Light Could Turbocharge Internet

ananyo writes "Twisty beams of light could boost the traffic-carrying capacity of the Internet, effectively adding new levels to the information superhighway, suggests new research. In the last few years, different groups of researchers have tried to encode information in the shape of light beams to ease congestion, using a property of light called orbital angular momentum. Currently, a straight beam of light is used to transmit Internet signals, but certain filters can twist it so that it corkscrews around with varying degrees of curliness as it travels. Previous experiments using this effect have found that differently shaped light beams tend to jumble together after less than a meter. Now, a team of researchers from Boston University in Massachusetts and the University of Southern California in Los Angeles has found a way to keep the different light beam shapes separated for a record 1.1 kilometers. The most imminent use of the cables, the authors say, might be to install them to span the short distances between servers on giant 'server farms', used by large Web companies such as Facebook."

Comments

I swear

[JMJimmy]

I saw a piece on this back in the early 90s on Daily Planet - I could never find it again and no one else seemed to remember it. Glad to know I'm not crazy!

Re:I swear

[gstoddart]

Glad to know I'm not crazy!

The one has nothing to do with the other ... you may correctly remember, but you might still be crazy. ;-)

Re:I swear

[Hillgiant]

I'm not crazy. You're the one who's crazy!

All I wanted was a Pepsi...

Re:I swear

[gstoddart]

Well done!

Wait, what are you talking about, WE decided!?
MY best interests?! How do you know what MY best interest is?

Re:I swear

[KingMotley]

Your memory is failing you. You read it back before you were transported back in time.

Re:I swear

[Anonymous Coward]

Back to when people used terms like "information superhighway?"

Awesome!

[Anonymous Coward]

More porn!

Re:Awesome!

[PPH]

More twisted porn!

Re:Awesome!

[hairyfeet]

Dude every time I think porn can't get more twisted i'll have a customer come in with a new porn bug and will completely break the scale again, so I'd say porn getting more twisted has got to be one of the laws of the universe.

Re:Awesome!

[Anonymous Coward]

I love Slashdot

Interesting idea, horrible post

[Anonymous Coward]

It's definitely a cool idea, but sheesh, this is Slashdot, people! We don't need a kindergartner's description of how the Internet and fiber optics work.

Re:Interesting idea, horrible post

[Sarten-X]

...But we do apparently need to be told what Hadoop is, or Drupal, or Ruby on Rails, or even SSH. There's always somebody complaining about too much simplification, and always somebody complaining about too little. Perhaps we could just learn to infer and ignore as appropriate for our level of prior knowledge?

Re:Interesting idea, horrible post

[Anonymous Coward]

Most people learn around age 5 that different people have different thoughts. Apparently not everyone learns that. You are right in that someone will always complain one way or another.

90ies called, want buzzwords back!

[i_ate_god]

It's 2013, does that make the term "Information Superhighway" retro?

Re:90ies called, want buzzwords back!

[jrms]

It's 2013, does that make the term "Information Superhighway" retro?

Judging from the summary, their idea sounds more like the Information Spaghetti Junction.

Re:90ies called, want buzzwords back!

[gstoddart]

Judging from the summary, their idea sounds more like the Information Spaghetti Junction.

Hmmm, if the internet is a series of tubes, then this is an intertwined set of bendy straws.

Re:90ies called, want buzzwords back!

[hendrikboom]

The internet is more like an ocean than a highway, anyway. Complete with surfers and pirates.

-- hendrik

Will this need new cables and other hardware

[Joe_Dragon]

If this needs new cables then maybe in the data center but installing them all over place will cost a lot + all the hardware that will need to be updated.

Re:Will this need new cables and other hardware

[Anonymous Coward]

Basically this has been done before, but it isn't very effective because it only retains its information carrying capacity for very short distances. What these guys have done is made the information coherent for over a kilometer. It could theoretically go for longer, but their cable was only made about a kilometer long. While interesting, it would require replacing cables rather than being useful for existing cables.

Re:Will this need new cables and other hardware

[Anonymous Coward]

Maybe apple should patent this.

Re:Will this need new cables and other hardware

[Charliemopps]

Not any longer. I work at a telco and a while back we got smarter. Rather than bury strait fiber, and have to dig it up when technology changes, we now bury flexible PVC tubing that the fiber rides in. When we need to change it, rather than dig it up at great expense, we find the particular fiber we want to replace, and just pull it out of the housing by hand. Then we use a device that shoots the new fiber through to the other side with compressed air. It's really neat to watch several miles of cable just pop out the other end like there was nothing to it.

I'm not sure if we used this stuff everywhere, and I'm not sure if every phone company used it. But it certainly saved us enough money that I'd imagine it's quite common now.

Re:Will this need new cables and other hardware

[similar_name]

From the summary:

The most imminent use of the cables, the authors say, might be to install them to span the short distances between servers on giant 'server farms', used by large Web companies such as Facebook."

Re:Will this need new cables and other hardware

[Gription]

It is kind of silly to focus on the need to switch cables since you would have swap the equipment that the cables plug into to get something capable of coding/decoding the information on each end.

Kind of like complaining, "But I'll have to get new shoe laces!" when you are buying shoes.

Re:Will this need new cables and other hardware

[Anonymous Coward]

Except in any case where the cable runs a distance more than just a single room, it becomes more and more difficult to replace the cable. It is not so much the cost of the cable as it is the effort to run new cable. If instead you could just get a better multiplexer that uses the same fiber, you could save some effort and money.

Re:Will this need new cables and other hardware

[jkflying]

Generally, the equipment on the end is the shoelace and the cable is the shoe. How much do you think it costs to lay a mile of fibre?

Re:Will this need new cables and other hardware

[Gription]

Generally, the equipment on the end is the shoelace and the cable is the shoe. How much do you think it costs to lay a mile of fibre?

From the RTFA dept...

The most imminent use of the cables, the authors say, might be to install them to span the short distances between servers on giant 'server farms', used by large Web companies such as Facebook.

Not germane to the subject seeing that I haven't seen a server farm that is more then a mile across...

Re:Will this need new cables and other hardware

[jkflying]

Which is precisely why it is being targeted at server farms: the cost of pulling up already-lain fibre in outdoor settings just isn't worth it, even with this additional capacity available.

Nothing new

[Anonymous Coward]

I read about them using angular momentum on wireless networks last year. I believe they did 2.5b/s on wireless.

Re:Nothing new

[Anonymous Coward]

Two and a half bits per second? Outrageous!!

Super computers

[Anonymous Coward]

In previous posts, some said the biggest limit for HPC is the communication bandwidth between nodes.

Exaflops might not be that far as away.

worst description of polarization ever

[Anonymous Coward]

Carl Sagan is turning in his grave right now... Or should I say 'corkscrewing'?

Re:worst description of polarization ever

[Sarten-X]

It's the "worst description of polarization ever" because it's apparently not polarization:

The orbital angular momentum of light (OAM) is the component of angular momentum of a light beam that is dependent on the field spatial distribution, and not on the polarization.

Re:worst description of polarization ever

[Anonymous Coward]

fail...

orbital angular momentum has nothing to do with polarization...

Re:worst description of polarization ever

[Technician]

It does sound like right or left hand circular polarisation to me too.

Re:worst description of polarization ever

[Anonymous Coward]

A beam of light that just has left or right polarization would be a beam that has that polarization everywhere in the beam. OAM means it has a spatial variation to the polarization, so that the polarization at each point in the cross-section of the beam varies. The name comes from the similarity to orbital angular momentum of a particle, which comes from it moving around at different points, as opposed to the spin inherent in the particle itself.

Re:worst description of polarization ever

[Ungrounded Lightning]

I think of it as being analogous to injecting separate beams of light at different angles, having them bounce back-and-forth between the walls at different distances between bounces, and emerge at angles corresponding to the angles at which they entered.

Of course it's not angle of flight that's in question, but another property of the light propagation that can be varied to allow different beams to propagate down the fiber and be separable at the far end. But they're still separate because each beam's cross section at a given plane cutting the fiber has a different distribution of phase and intensity, resulting in different propagation mechanisms that conserve a property which can be used to separate the beams when they emerge.

Re:worst description of polarization ever

[__aaltlg1547]

What it means is that they're using other than the TEM mode of multimode fiber. It can be done but it has problems with dispersion and decoherence with distance. Every optical and probably ever electromagnetics engineer learned about this in school and most of us concluded that single-mode fiber was preferable because it doesn't have those issues.

The standard method for multiplexing signals on fiber is therefore to put it on different colors instead of into different excitation modes. That way they can be used over very long distances and there's zero bleed-over from one channel to another.

Re:worst description of polarization ever

[__aaltlg1547]

More or less. More detail here: http://stanwir.seecs.nust.edu.pk/Lectures/FOCS/Modes%20in%20Optical%20Fibers.pdf

It pains me to say this, but

[slew]

People are probably beter off reading the wiki...

Key bit of information...

OAM multiplexing can not be implemented in the existing long-haul optical fiber systems, since these systems are based on single-mode fibers, which inherently do not support OAM states of light. Instead, few-mode or multi-mode fibers need to be used. Additional problem for OAM multiplexing implementation is caused by the mode coupling that is present in the fiber, making direct-detection OAM multiplexing still not being realized in long-haul communications. In some specialty fibers, OAM states were transmitted with 97% purity after 20 meters.

Basically this demonstration technique uses specially designed fibers that can carry the "donut" TEM mode required for OAM which is the reason they made a comment that the most likely for fibers the implement this technique "might be to install them to span the short distances between servers on giant 'server farms'"...

Re:It pains me to say this, but

[Guspaz]

Is there actually any capacity crunch on the physical fiber itself that makes it worth replacing? My understanding is that existing DWDM hardware can pump multiple terabits per second through a single fiber strand, and that the reason pumping a terabit through fiber requires multiple transceivers is simply because the electronics can't drive data rates, not because of any limitation of the fiber...

It seems that OAM would have the same problem; you still need electronics that don't exist if you want to push that much data through any single medium, and we're not even remotely close to the limits of regular fiber, so does OAM actually enable anything?

Re:It pains me to say this, but

[Shatrat]

You've hit on the real reason nobody is interested in OAM/Spatial mutliplexing. Depending on the vendor, we can light 80 to 160 channels of DWDM today on a pair of fibers, and go for thousands of kilometers in a well planned system.

Re:It pains me to say this, but

[freeze128]

...or we could just different colored lights in the existing fiber....

Re:It pains me to say this, but

[pv2b]

This is what WDM (Wavelength Division Multiplexing) means. (DWDM = Dense Wavelength Division Multiplexing).

Re:It pains me to say this, but

[Shatrat]

What do you think DWDM is?

Re:It pains me to say this, but

[Anonymous Coward]

That's what WDM does.

What I think you're actually suggesting is that different parts of the IR band (1200-1700 nm) be used supporting thousands of DWDM channels, and they can on modern fiber. The problem is that most fiber, the deployed, dark, and effectively free fiber left from the bankruptcy of the dotcoms, has a very narrow bandpass.

When we run out of "free" fiber "different colors" will be deployed.

Re:It pains me to say this, but

[Anonymous Coward]

Is there actually any capacity crunch on the physical fiber itself that makes it worth replacing?

Increasing capacity on existing long-distance fiber even a little bit definitely has huge value. If the fiber has to be replaced, sure it could still have value iff it's cheaper per bit carried, not just more bits per fiber for much more $, though I speculate this is much less valuable than lasers or optics for existing fiber.

However, they're offering something completely different: <1km. That does not have value. We already run systems a few km long single-color or CWDM (8 colors) because it's cheaper to buy more fibers than get fancy DWDM (64 - 80 colors) lasers and splitters. Sometimes we even do EOE in datacenters because fibers are cheaper than fiber connectors. If you use an electrical termination it can make sense to do 2.5Gbit/s per fiber instead of 10Gbit/s per fiber just to save money and energy on the MUX chip. Datacenters care about watts per port, cost per port, connector and laser cost, cable managability, but they just don't care about bits per glass fiber. It is the thing that matters least.

How does this work

[Anonymous Coward]

How the hell does this make the "Information Superhighway" faster? I'd think that making the light travel farther would cause it to slow down transmission speeds.

Re:How does this work

[Anonymous Coward]

Bandwidth is a ratio of data size to speed. (Data Size / Time) If data size increases by a lot, but the speed slows down a little, it is a net bandwidth is increased.

The example of having a truck loaded up with hard drives driving down the highway has a larger bandwidth than many other options.

Re:How does this work

[Anonymous Coward]

The light "twisting" happens even in a vacuum, so the light isn't actually taking a longer path. The twisting is a property of the photons and an interaction among the group of photons.

Any extra latency added will be from the minor doping of the fiber to cause the different "shapes" to have ever so slightly different refraction values. The actual speed of the light, end to end, should be quite similar.

Is color used?

[Anonymous Coward]

Btw, is color of the light used yet?

Naive, top of the mind idea: If bit is 1, and used "red channel", and if bit is 1 and using "blue channel" same time then the color would be purple.

Re:Is color used?

[PPH]

Yes, color is used. But not the way you think it is.

Chromatic dispersion causes different wavelengths (colors) to propagate at different speeds. So what would start out as a coincident red and blue bit wouldn't stay that way very long. But using separate colors as one would use separate fibers with each carrying its own data stream does work.

then the color would be purple.

Your eyes and brain work that way. Other sensors don't. A 'red bit' plus a 'blue bit' will not trigger a purple sensor.

Re:Is color used?

[Anonymous Coward]

Colors are varied wavelengths, which would be used in muxing signals together to increase bandwidth. I can not say from experience, but I would assume it is being used.

Improving signal retention (stopping degradation if there is any) to allow improvements in reading signals would work too, but this is getting harder and harder.

Re:Is color used?

[Shatrat]

What you're describing would be WDM, and this is OAM or Spatial Multiplexing depending on who you talk to.

Your eyes and brain work that way. Other sensors don't. A 'red bit' plus a 'blue bit' will not trigger a purple sensor.

That's not really true, photodiodes and avalanche photodiodes are not channel specific. I can connect a transmit from a 1560.61 to a RX of a 1558.98 and they will work just fine. Most are wide-band and will happily receive anything from 1250 to 1650 nanometers.
If I mix those wavelengths with a combiner and send them into a receiver, I'll get a loss of frame because the signals conflict. The channels have to be 'demulitplexed' at the far end with some kind of WSS, AWG, FBG or other optical device.

Re:Is color used?

[Anonymous Coward]

Your eyes and brain work that way. Other sensors don't. A 'red bit' plus a 'blue bit' will not trigger a purple sensor.

Different sensors may have different sensitivity at different places in the spectrum, but there are plenty that can cross the whole visible spectrum plus more. It would be easy enough to make a detector that responds to either (or requires both) with the appropriate notch filters in front of it. Of course a violet filtered sensor won't detect blue or red... but a filter for purple = red + blue and an appropriate threshold would (or two sensors and some logic for more robustness).

Who'll patent it first?

[penglust]

Since this was created at a University it is unclear on which company will attempt to patent it first. Any bets? Should be start a pool?

Re:Who'll patent it first?

[Anonymous Psychopath]

Since this was created at a University it is unclear on which company will attempt to patent it first. Any bets? Should be start a pool?

Why wouldn't the university patent this? That would be what normally happens.

Re:Who'll patent it first?

[TheSkepticalOptimist]

My bet is Apple will try. but it doesn't matter who patents it now because the university already holds prior art due to having published the scientific paper first. There is no rule that a university cannot hold a patent, the only shitty thing is if a student was the one that actually invented it, it's his professor that will take credit as the inventor. That's why you wait till you graduate before you invent shit.

How useful is this, really?

[sirwired]

If they ever got the distance extended, I could see this being handy for long-haul links (although is it really better than a CWDM or DWDM?) But over intra-rack links, it'd have to have a pretty small premium over current technologies to be able to justify using it over just laying another strand of fiber.

In all but HPC applications bandwidth is rarely an issue anyway. You might occasionally use those for trunk lines, but again, why not just lay another strand?

Re:How useful is this, really?

[Shatrat]

Laying another cable is crushingly expensive, but you're right this technique is still so vastly inferior to WDM that nobody is seriously pursuing it outside of science projects.

Finally, a comeback for Monster Cables!

[Anonymous Coward]

Luxophiles will appreciate the DEPTH of the new shaped light...

Re:Finally, a comeback for Monster Cables!

[rubycodez]

they have similar marketing spew already for their overpriced fiber optics:

http://www.monstercable.com/productdisplay.asp?pin=3792

There's something very important I forgot to tell

[Anonymous Coward]

Peter Venkman: What?
Spengler: Don't cross the streams.
Venkman: Why?
Spengler: It would be bad.
Venkman: I'm fuzzy on the whole good/bad thing. What do you mean, "bad"?
Spengler: Try to imagine all life as you know it stopping instantaneously and every molecule in your body exploding at the speed of light.
Ray Stantz: Total protonic reversal.
Venkman: Right. That's bad. Okay. All right. Important safety tip. Thanks, Egon.

Wouldn't light have to travel longer distance?

[Anonymous Coward]

With having to travel "twisted" pathway instead of straight-through, wouldn't light have to travel longer distance, hence increasing latency? Server Farms aside, ability to transfer x2 the amount of information over the long distances is going to become rather irrelevant since your latency almost doubles.

Re:Wouldn't light have to travel longer distance?

[Githaron]

From the article:

The work published today used clockwise and anticlockwise versions of twisted light with a specific curliness, but Ramachandran says that the team has since done other research that suggests that about ten different beam shapes can be used to convey information.

That is exciting because each shape could potentially act as an entirely new level of traffic on the information superhighway. On each level, streams of data could be further divided into narrow lanes of colour, maximizing flow. "We showed a new degree of freedom in which we could transmit information," says Ramachandran.

It sounds like you are getting ten times the information. Even if latency doubled, it might be worth the tradeoff with applications such as file transfer and music/video streaming.

Re:Wouldn't light have to travel longer distance?

[Anonymous Coward]

The light does not actually "twist" in a physical motion, just a logical concept. And it's not 2x the bandwidth, it's 10x.

This tech should be re-billed

[Anonymous Coward]

The major cost of fiber is in it's installation rather than materials, meaning that laying 192 strands of fiber for X miles might cost 5% more than laying 48 strands of fiber. So there's not much of need to pack more data on a single fiber because there's little extra cost associated with using parallel paths, and that's not likely to change.

Can not be used on existing fiber

[silas_moeckel]

Do they realize how much fiber is lit at gigabit and below? 10ge is hitting mainstream and 40 and 100 following on quickly. CWDM is dirt cheap and DWDM is getting there as well. Replacing all the long haul fiber is a non starter could be useful in 100ge short reach dropping it from 10 or 4 lanes to 2.

One the physical side 25gbps looks to be the plateau we are hitting 16 lanes gets 400ge ports.

bad move reference...

[wbr1]

Curve the bullet.

I don't understand

[Anonymous Coward]

How corkscrewing light will turbocharge anything...

Will the exhaust light impact an impeller that then compresses the light on the other side?

Because (I direct this to every idiot who mis-uses the term "turbocharge") THAT is what a TURBOCHARGER is.
A compressor that is powered by exhaust gas.

If whatever damn device you are speeding up does not contain:
an waste exhaust-driven impeller
an impeller driven compressor
then it is NOT turbocharging. /end of rant

Re:I don't understand

[rubycodez]

fine, we'll twincharge and supercharge and hybridcharge our CPU from now on

Re:I don't understand

[Iniamyen]

Every child of the 80's knows that "turbo" just means "better." Move along...

OH OH

[Nov8tr]

We're all "screwed" now.......grin :P

Would Quantum Communications be better?

[Anonymous Coward]

Thoughts, riddle me this one, if Quantum Communication finally makes it to manufacturing with network interfaces for each end, then this light stuff is obsolete!

Thank you Uncle Albert Einstein!

Listen up, twisted light is fine, but Quantum Communications is mighty fine...

use in conjunction with WDM maybe?

[Chirs]

It seems at least plausible that this could be used in conjunction with some form of WDM to get higher densities than either alone.

Works Great

[sexconker]

Works great as long as all your cables are perfectly straight. If you want to bend the cable, you need to plug it into an angled repeater.
The repeater, of course, requires an external power source.

Re:Works Great

[Anonymous Coward]

That was the old problem. They have fixed that problem by doping the fiber. You can now bend the fiber.

how is this different from this story in 2012?

[AndroSyn]

This sounds like what these guys were doing: http://hardware.slashdot.org/story/12/06/25/1215246/twisted-oam-beams-carry-25-terabits-per-second

Not sure if its the same groups or not, but pretty much the same idea.

Not good for me

[Anonymous Coward]

since I use the Internet to unwind

what next?

[Anonymous Coward]

...So now the internet comprises of cats, pipes and now corkscrews!?

solving a non-problem

[__aaltlg1547]

This is just a dumb idea and a waste of investors' money. The technology already exists to put many colors on the same fiber and unlike optical modes (which is what they're really talking about) colors don't change over distance or bleed into one another.

Not what I expected...

[Anonymous Coward]

I opened the link fully expecting to see data cables routed through those annoying CFL bulbs, with the same claims: "56kbps of corkscrew data transmission is equivalent to 100Mbps of conventional data transmission...and it uses less power! ...and the government is going to push it by outlawing 100Mbps connections!"