Slashdot Mirror
Scientists Double Optical Fiber Transmission Capacity
ms writes: "Yesterday golem.de reported that the Optical Communication and High-Frequency Engineering Group at the University of Paderborn (Germany) claims to have made a technology practical which doubles the transmission capacity of optical fibers to 80 GBit/s. In their so-called "polarization division multiplex data transmission system" they don't only send one but two mutually orthogonal light waves through the fiber. They say the only big problem was the dispersal of the light waves which limits the data rate. Additional they had to fight against the phenomena that the polarization direction of the light waves changes while it goes through the fiber. Now, after two years of research, they invented an "automatic optical compensator of polarization mode dispersion" which fights both the limitations. With this gadget they were able to send data at a rate of twice 40 GBit/s (that's 85,899,345,920 Bps) over a test-line of 212 km. And "only the available equipment limited distance and data rate". As we all know, optical fibers build the (cronically overloaded) backbone of our beloved Net. (BTW: That's Net., not .Net!)" Here's the babelfish translation, too.
Just trying to grok "mutually orthogonal". Is that redundant, or just over my head? Not trying to nitpick, but to understand something my networking prof never explained.
"What is the sound of one belly slapping?"
-Foxxz
Additionally it possesses a controlling mean, which is to after-pursue even largest polarization modifications, as they occur on very long transmission circuits contrary to competitive systems also, noly-break.
Once I figure out what a noly-break is, I should be able to build my very own high speed home network!
I Am My Own Worst Enemy
Maybe some day cutting a fiber line would yield deadly results. Imagine accidently digging through a fiber line only to be cut in half by the power of the beam. That would rule! As if being electrocuted to death wasn't enough, soon we can be killed by data...I can see where the 'freak accident causes supergenius to be born' movies are going to come from..
Job? I don't have time to get a job! Who will sit around and bitch about being broke and unemployed then?
It's ironic that this article comes on the heels of some articles I just saw (in metamoderation) about how oversold services (DSL/Cablemodem) are. This would be a great thing, if any company still has the capital to lay new fiber, though I expect many are just trying to eke out a returns on what they already have. My ISP, Concentric/XO was last listed at $0.40 a share, and they're advertising cheep long distance, etc. right now.
A feeling of having made the same mistake before: Deja Foobar
It seems every year we find a way to double the amount of data that we can send down fiber. As a result of this, companies are actually deploying less new fiber in the field and taking older lines out of commission.
One of the things that worries me about this is the increased vulnerability. In the past, huge fiber networks were used that can be one tenth the size today. All too often a clueless construction worker rips up a section of fiber and causes some havok.
Won't this kind of thing happen more frequently if less fiber is deployed that can handle more traffic? And does this bring us any closer to fiber to the curb - it doesn't seem like it.
First, twice 40Gbps is not 85,899,345,920 bps, it is actually 80,000,000,000 +/- 1,000,000,000. We don't measure Gbps in powers of 2. Secondly, the internet backbone is not overloaded, but is running at about 20% capacity according to the people who operate it.
Hmm, and same Timothy posted this article on June 25th about a lot of fiberoptic cables that have been put into the ground but haven't been put to work. :)
You gotta love the consistency of Slashdot posts
"A door is what a dog is perpetually on the wrong side of" - Ogden Nash
In geometry, orthogonal just means perpendicular. But, according to searchStorage: "In computer terminology, something - such as a programming language or a data object - is orthogonal if it can be used without consideration as to how its use will affect something else. " So, the light waves are mutually orthogonal (they are data objects in this case), but I'm not exactly sure how to apply the definition to exactly what the scientists are doing with fiber optic cables.
F-bacher
James Tiberius Kirk: "Spock, the women on your planet are logical. No other planet in the galaxy can make that claim."
"mutually orthogonal" means (for a set of two or more elements) that each pair of elements is orthogonal--AFAIK, it's a synonym for "pairwise orthogonal". "orthogonal," of course, has lots of synonyms, including "linear independence," "at right angles," "having zero dot-product," "statistically uncorrelated," etc.
So, the three spacial dimensions, the set {phase of the moon, day of the week, time of day}, etc. are all "mutually orthogonal." When talking about a set of only two elements, the "mutually" is superfluous, but not redundant.
-- MarkusQ
Imagine if HTML was patented way back in the 1.0 days. By now the owners of the patent would have:
Instead they gave freely and changed the world. ;)
"What is the sound of one belly slapping?"
The beams in this article are orthogonal in the sense that channel #1 has it's E-field pointed prependicular to channel #2's E-field so they won't interfere with each other (so they're `orthogonal' in the usual compu-geek sense of the term, too.)
The german team seems to have solved two big engineering problems with sending two channels of information this way. One is to send a mean-polarized signal so that you can compare the two channels against it (kind of a carrier signal for polarization) to see which channel is which.
The other I confess to not understanding. Apparently there are sync problems -- signals carried one polarization may travel faster than the other polarization. I can only guess that this is a problem caused by inhomogenaities in fibre. Whatever its caused by, they've managed to measure it and compensate for it.
As for your other question, they definately can and do use frequency as a way of encoding information. Just like with radio signals, you can use the brightness of the light (amplitude modulation, or AM) or its color (frequency modulation, FM). In practice, FM is less problematic; the amplitude of a signal is easily confused by noise, whereas frequency is much less so.
We already have much more fiber capacity than we can use. The real bottleneck of the Internet right now is... the switching. OC-768 units (38.8 gigabits/sec) won't even reach volume production until 2003 or later, and they wouldn't even handle half of one of these fibers, let alone multiple fibers coming from various locations. It's like running a 2" diameter fuel line to the engine of your Hyundai.
All-optical switches have been developed, but are not going to be widely deployed for some time. I have a feeling that even all-optical switches will be many years before they reach the speeds needed for 80 gb/s fibers.
The true improvement of the Internet will occur when switching capacity increases by at least an order of magnitude in a very short amount of time. Right now, good, guaranteed bandwidth is barely any less than it was back in 1997. Sure, as switching capacity slowly progresses to fill the needs of the backbone providers, the Internet keeps running - but you still end up paying out the nose for guaranteed bandwidth. Once the switches catch up with the fibers, however, that *might* change. Maybe.
steve
Oh, you're not stuck, you're just unable to let go of the onion rings.
I propose that in response to Microsoft's usurping of the dot, we counter-respond with an acquisition of the asterisk.
.NET, or The Net, or 'net, or Net, we all start using *Net.
Thus, instead of using
That way we can pretty much cover all bases, since everyone knows that * is a wildcard - the All-unifying Infinite Eternal Symbol of All.
All those in favor, say "*me"...
; -- the corruption of government starts with its secrets. a truly free people keep no secrets. --
I know it ain't easy. You try to be a troll, and some asshole moderator gives you (+1, Funny)...
I don't understand the importance of this discovery. I'm pretty sure existing DWDM systems can put at least 16 wavelengths on a single fiber at OC-192 (10 Gbps) speeds for a total capacity of 160 Gbps...
This is kind of an intresting experiemnt, but this is not news. The "otherwise possibe" part makes it sound like no one has done PMD compensation before, this is false. Here is why:
1. PMD compensators are being built by many research groups. You still can't call up an order one (AFAIK), but soon.
2. PMD (mean DGD, differential group delay). DGD changes with time and wavelength.
3. PMD on buried fiber varies slowly. It is easy to compensate.
4. Nortel, Alcatel and others with be releasing 40 GB/s (per WAVELENGTH) systems next year. They are suppose to run 100's of km, between regens and at many wavelengths (160?).
Here is a link with almost all peer reviewed papers on PMD:
Here one can see many references to PMD compensation and even some at bit rates of 160 GB/s. With PMD compensation the line speed isn't that important, it is the accuracy and speed of your compensation.
This is not a break through.
Good post. I agree with you, and naturally disagree with the "we must share all advances for the good of the people" attitude. Communism isn't counter-culture, anti-establishment, nor is it cool. It just plain sucks.
Good post for you too. It is one thing when people choose to share things for the advancement of all people, like GPL and public domain and such, but it is a different thing when you don't believe in property rights at all.
I've had enough abrasive sigs. Kittens are cute and fuzzy.
Orthogonality in this case refers to modes of the electric field. A mode is labeled by a frequency and a polarization state. So, yes, two beams of different wavelength are orthogonal modes, and can be resolved by using a grating to diffract each component to a seperate detector. This is called WDM -- wavelength division multiplexing.
States with opposite polarization (horizontal and vertical, right and left circular, etc) are also orthogonal modes, and can be seperated, for instance, by a polarizing beam splitter. This is PDM.
The total bandwidth of a communication link is bps/mode * useful modes, so either increasing the number of frequencies or polarizations, or both, can improve bandwidth.
WDM is limited because each pulse actually covers a range of frequencies, and you need to choose them far enough apart that they don't overlap, or you won't be able to discriminate them well. PDM is limited because it is hard to get fibers to not fs*k with the polarization of light, plus there are only two orthogonal states, so you can only easily get a factor of 2 improvment in bandwidth.
I see some others posting explanations about physics behind this, but it seems a bit unsatisfactory for some. Here's my best shot at it:
There are two orthogonal polarization modes that propagate down fiber, meaning the there's a sort of up-down oscillation of the electric field (one mode), and a left-right oscillation (other mode). If fiber were perfect, you could send a signal along each polarization, and they wouldn't bother (interfere with) one another, but it's not. If you send polarized light down a fiber, it will not keep the same polarization (unless you use polarization-maintaining fiber, but that's a pain, and you can only send one polarization down).
So people generally send down (relavitively) unpolarized light. They modulate this one signal as fast as they can (getting about 40Gb/s), and then deal with dispersion as best they can.
Dispersion results from the spread in frequencies (colours) of your signal (each colour travels a different speed in the fiber) and also from the fact that a fiber has polarization mode dispersion (the part of the signal along one polarization axis travels at a different speed than the other part, called PMD from here on in). Both of these effects cause a pulse that you send down the fiber to be distorted (part of the pulse travels at a different speed than the other part). Chromatic dispersion (the first kind) has been dealt with (fibers have a wavelength at which the loss is lowest and a wavelength at which the chromatic dispersion is lowest, and it's been worked such that these two things are at basically the same wavelength), but PMD is a big limitation to pushing the capabilities of fiber. This was stated on the front page post:
They say the only big problem was the dispersal of the light waves which limits the data rate.
I think that should read "dispersion", not "dispersal".
So, what these guys have done is made a PMD compensator. Somehow it automatically makes sure that a given polarization of light stays in that polarization as it travels down the fiber. If one can preserve the polarization of both modes (which is different than polarization maintaining fiber, which takes ONE polarization of light and keeps it polarized), and then send a signal along each polarization axis, then one doesn't need to deal with PMD, because within a given signal, all the pulses are travelling at the same rate.
Then, if you don't have to deal with PMD, then there's very little to slow you down in pushing data through the fiber, basically just how fast you can modulate your laser (I think you could drive a LiNbO3 Mach-Zhender modulator up to about 80Gb/s or so, whereas I think in the article they were driving it at 40Gb/s). That's why they say the data rate was only limited by available equipment. I'm not sure how the PMD compensator works, I'll have to read the actual article more closely. I hope this helps!
"Now gluttony and exploitation serves eight!" - TV's Frank
Making bigger and bigger pipes is the way of the net. The technology just has to get better by making the rates higher and the packages smaller.
The next question is, what do we do with all this new capacity? The telecommunications industry is finding all this out since so much infrastructure was built during the boom and everything went bust. There was so much infrastructure out there that was completely useless (think dark fiber) because of incomplete implementation.
They put plenty of fiber in the ground and run out of money before they can get any equipment to light it up. They put plenty of DWDM equipment to light the fiber but they can't sell a whole OC-48 to anyone at a competitive price. They get tons of metro networking equipment but they've just spent all their money and can't make a management system to turn up any metro circuits.
So, these kinds of technological improvements are the greatest thing in the world and they have to keep pushing the envelope. However, there must be concurrent development along all other lines to manage all these ass-kicking boxes and make them usable! If not, no one will buy it because it can't contribute to a sustainable business model.
"Money often costs too much" -- Emerson
In WDM (I'm not sure the acronym is right) what you do is send more than one signal by using different light colours (frequencies) that don't interfer with each other.
What you do here is send two signals that have the same frequency but orthogonal polarizations so they don't interfere to each other either; imagine one of the waves going on the vertical direction while the other goes horitzontal, that would be orthogonal and you could recover both the vertical and the orthogonal one independently.
We did a test on a private WAN and about 50% of the traffic was heartbeats.
- - - - - - - - - - -
I am a programmer. I am paid to produce syntax not grammar. Deal with it.
Interestingly, atleast to me, that's almost exactly 4x per year. As in 300 * 4^14 is almost exactly 85,899,345,920.
Justin Dubs
DWDM (dense wavelength division multiplexing) referrs to multiplexing multiple optical signals on a fiber by having them exist at different wavelengths of light. This is very similar to how the cable TV line carries 100 or so channels of TV signal by having them at different frequencies.
The D (for dense) means that there are many such channels, often 40+. This article referrs to having two 40Gb/s channels at the same wavelength, but with opposite polarizations so they don't interfere with each other much. This same signal could be used as a base for a DWDM system to effectively double the current maximum speed of like 10Tb/s (40Gb/s * 250 channels).
main(O){10<putchar((O--,102-((O&4)*16| (31&60>>5*(O&3)))))&&main(2+ O);}
LN2 is cool!
This phenomenon is called polarization-mode dispersion and we just covered it in my fiber-optic communications class. It occurs because of birefrigence, which is the phenomenon where different polarizations see different refractive indices. Since refractive index is the speed of light in vacuum divided by speed of light in a medium, this means signals with different polarizations will travel different speeds. Even worse, since fiber birefringence is probably stress-induced and varies over the length of the fiber, it is difficult to tell what the polarization axes of the fiber are so that you can minimize this effect.
Polarization-mode dispersion is a problem even when you're not multiplexing by polarization because it results in the ordinary and extraordinary polarization of a light pulse separating and possibly colliding with other pulses, thereby limiting the bandwidth of the fiber. On the other hand, if you use the ordinary polarization as one channel and the extraordinary polarization as a separate channel, both channels will propagate with zero polarization-mode dispersion and double the effective bandwidth of the fiber. They will propagate at different speeds, but that really isn't an issue as long as the light pulses that represent your 0's and 1's aren't spreading.
The trick is determining the ordinary and extraordinary axes of the fiber, which is the breakthrough that this group made. It sounds like they use a reference channel to determine the ordinary and extraordinary polarization axes of the fiber and also to measure the change in polarization introduced by the fiber so that they can demultiplex the two polarization channels. This is a very simple and elegant way to negate polarization mode dispersion and to enable polarization-division multiplexing.
"It take 9 months to bear a child, no matter how many women you assign to the job."
- As we all know, optical fibers build the (cronically overloaded) backbone of our beloved Net.
If it's overloaded by cron, couldn't we just kill the cron daemon?I/O Error G-17: Aborting Installation
Well, that 300bps was what you could manage across an analogue telephone line with a commercially-available modem.
Telecom backbones, even 14 years ago, were waaay beyond 300bps. And this technology really is backbone stuff, not for us consumers *wistful sigh*
You folks are using NetBeui as your filesharing system aren't ya?
Switch it to NetBeui over TCP/IP (if you wish to avoid NFS) and you'll cut the heartbeats to 1/10th.
Microsoft products and NetBeui are very chatty with the 'are you still there?' crap.
Rod Taylor
If you could make the cable dense enough to fit in a space the size of a normal harddrive -- and made it low power (already solid state) that would rock my world as a swap device.
I'm willing to accept an average of 1msec seeks (600km of fibre) to double capacity.
Rod Taylor
The fact that they've just doubled it to 160GBps in your case.
It's not DWDM that they're using -- but (from what I can tell) it also doesn't negate using DWDM along side.
Kinda like satelite transmissions using horizonal, vertical, left circular and right circular polorizations at the same frequency. Then for kicks they start using many channels in that manner.
Rod Taylor
Best you can get in openair is about 4 polorizations at any given frequency.
horizonal, vertical, left circular, and right circular. They're a bitch to collect all 4 at the receiving end though.
I doubt that fibre could handle the circular modes -- but I'm not a fibre person.
Rod Taylor
I know this is a little off topic, but I think the correct term is "Backhoe Fade". A quick search turned up an "official" government project .
BTW, the word "fade" is a throwback to the time when most longhaul communications was done using troposcatter microwave systems. Small atmospheric changes such as rain, volcanic dust, solar flares and sunspots would cause the Recieve Signal Level to drop. Ok, Ok, who am I kidding; those bastards would fade at dawn, dusk, mid-day, mid-night, summer and winter solstace, equinox, and any time someone stood too close to the radio. A guy I knew actually shot a radio once for excessive fade. He claimed it was an accident, but the investigator was clued in by the fact that there were 3 holes in the radio...
I'd rather you do it wrong, than for me to have to do it at all.
Bytes per second != bits per second
Somebody please correct the story.
echo '[q]sa[ln0=aln80~Psnlbx]16isb572CCB9AE9DB03273snlbxq' |dc
"polarization division multiplex data transmission system" [using an] "automatic optical compensator of polarization mode dispersion"
If it's twice as fast but takes four times as long to say it, does that actually mean its effect is half the speed? The article didn't say if the "test-line of 212 km" was just so they could write the name on the side.
And to think people believe we IT staff make up impenetrable terminology in an attempt to justify our salaries!
Communications engineers deal with bits or symbols, not bytes. In that context, kilo and mega are almost always the normal SI meanings (10^n).
Mea navis aericumbens anguillis abundat
While I am always for extending bandwidth, I jump on the term "chronically overloaded" (note the letter "h", which seperates chronic from mere cronjobs)... Not long ago, regarding the 9/11 effects of everyone and their dog logging on to the net on the search for information, I read that the net backbone currently is anywhere between 40% and 50% at any given time. It took 9/11 to get peaks up to 80%-85%.
I tend to think being at half load most of the time is not anywhere near "chronical overload".
+++ath0
No, 2 is the max. Circular polarisation is just horizontal AND vertical at the same time, with the same amplitude and 90 out of phase. So you will not be able to discriminate the channels if you add in extra horizonal and vertically polarised light as well.
And yes, fibre can handle circular modes, or any other polarisation state for that matter.
Last geeky point. Orthogonal can be rephrased as '2 polarisation states occupying antipodean points on the Poincaré Sphere'
-- MarkusQ
in Newark, NJ. Nobody was using it for data back then of course, but squeezing 1.544 megabits/sec onto a copper pair, in order to move voice circuits around, is nothing new.
I wish the Slashdot team knew a bit more about telecom, they'd accept fewer of these stories and say fewer stupid things about them. "Chronically overloaded backbone" my ass, there are millions of miles of dark fiber out there. The glass isn't the problem, it's the silicon, the greenbacks, and the red tape that make things suck.
I always wondered what the fuck circular polarization was. So you don't really need those curlicue antennae, you could just make two antennae, orient them 90 degrees from each other, and then set one a quarter wavelength behind the other, right?
I don't personally have any experience with Erbium-doped fiber amplifiers, but someone here must...
Do EDFA's clobber polarization or not?