UCL Scientists Push 1.125Tbps Through a Single Coherent Optical Receiver

Mark.JUK writes: A team of researchers working in the Optical Networks Group at the University College London in England claim to have achieved the "greatest information rate ever recorded using a single [coherent optical] receiver", which was able to handle a record data speed of 1.125 Terabits per second (Tbps). The result, which required a 15 sub-carrier 8GBd DP-256QAM super-channel (15 channels of data) and total bandwidth of 121.5GHz, represents an increase of 12.5% relative to the previous record (1Tbps). Now they just need to test it using some long fibre optic cable because optical signals tend to become distorted when they travel over thousands of kilometers.

Comments

Big deal

[110010001000]

I was able to do that in Linux with a few shell and Perl scripts.

Re:Big deal

[Etherwalk]

I was able to do that in Linux with a few shell and Perl scripts.

Ah, the speed and sophistication of interpreted languages.

Re:Big deal

[ickleberry]

I was able to do that with a slightly overclocked PIC12F683 and a few lines of assembly

Re:Big deal

[maestroX]

You .. modem hoarder!

Re:Big deal

[GinRummy33]

Wait. they got 1.125 tablespoon of what?

Re:Big deal

[peragrin]

Do you know how hard it is to get a tablespoon of light to stay in place. It just wants to go everywhere.

Re:Big deal

[Anonymous Coward]

Trap that light in a tablespoon of honey and you're golden.

Terabit ethernet

[Anonymous Coward]

Could this be used for terabit ethernet? Many companies including Facebook and Google have indicated the need for terabit ethernet in their data centers. I'm surprised there isn't much effort to develop terabit ethernet. It sure seems like this would be useful for it, though.

Re:Terabit ethernet

[Shatrat]

This is aimed at optical transport. The framing will be OTN and it can carry many payload channels of 10GBE, 100GBE, or Sonet, SDH, whatever is needed. Ethernet sucks for long distance transport because it doesn't have built in layer one performance monitoring to match OTN and even old school Sonet/SDH.
Right now you can buy a single transceiver from Infinera that will do 500gbps using 10 carrier wavelengths. Ciena, Nokia and some others offer 200gbps over short distances on a single carrier. So, 1.125 gbps over 15 carriers isn't a huge leap forward, but is going to be table stakes for the next generation of optical transport.

Re:Terabit ethernet

[Anonymous Coward]

This is 256QAM, the same technology used in DOCSIS cable modems, so it is a digital signal modulated onto analog carriers. As such it really is not going to be anything you would use for long distance transport as you cannot regenerate analog signals without converting them back to digital to do so. There are some things you can do with dispersion compensation and the like, but really if this were to have any applications I would think it would be more in the FTTH area.

Re:Terabit ethernet

[Shatrat]

You don't have to regenerate the signal, just amplify it. Optical amplifiers have been around for decades and with current technology you can go thousands of kilometers without an OEO (optical-electrical-optical) regeneration just by placing amps every 100km or so. Chromatic dispersion is irrelevant these days because the Coherent optics have ridiculous CD tolerance.

Re:Terabit ethernet

[Woldscum]

Up to the late 90's @ a regen shelter you needed to WDMdemux-Regen-OPAmp-WDMmux. Late 90's Nortel came out with MOR (Multiple Wavelength Optical Repeater) cards. Which replaced the WDMs and wavelength specific transmit regen cards but still used OPAmps. The OPAmps of the day had a theoretical range of 90 miles (at least Pirelli and Siemens did). Depending on the condition and quality of the OSP fiber you also need DSMs (Dispersion Compensation Modules). Latter Siemens and Ciena systems today do Regen and Amplify all in one. You must regen. Just amplifying degrades the signal too much. You are also referring to subsea cable. Which is a totally different doped glass compared to the stuff used on land.

Re:Terabit ethernet

[Shatrat]

You're 3 decades out of date. You can amplify dozens of times and cover thousands of kilometers these days. Dallas to Chicago is a breeze. You can even optically express certain channels and drop others, and redirect them down different fiber paths dynamically based on impairments. Read up on hybrid EDFA/Raman amplifiers and flex-grid ROADM.

Bah, I was doing that last year

[Anonymous Coward]

I 3D printed the whole thing. Onlty Luddites wait for others!

For Reference

[Anonymous Coward]

My first modem was a blazing 300 baud so in a moment of nostalgia I figured out how long it would take my old modem to get the equivalent of 1 second's worth of this optical receiver's speed.

1.125 Tb / 300 b = 3.75 billion
3.75 billion seconds / 86,400 seconds / day = 43403 days
43403 days / 365 days/ year = 119 years.

Obligatory "I can now transfer my porn collection in mere HOURS with this wonderful technology"

Re:For Reference

[sycodon]

And Windows still take 30 seconds to display the contents of a folder.

What about the incoherent optical receiver?

[mveloso]

So, what kind of speed do they get with the incoherent optical receivers? 1 kb/sec?

Oh, so it talks?

[Marrow]

This is coherent light.

Just like Up tends to be Up and Down Tends to be..

[Anonymous Coward]

> Now they just need to test it using some long fibre optic cable because optical signals tend to become distorted when they travel over thousands of kilometers.

How about::

'While this represents a new record for short range transmission, the measurement of optical communications depends on the distance over which a signal can be transported. In fiber optical communications, the"bandwidth-distance" metric is limited by fundamental interactions of light with the fiber. The maximum "bandwidth-distance" achievable is on the order of a few "THz-KM"

UCL has an Electronics school?

[monkeyxpress]

I love London, but it surprises me that anyone would come here to study Electronics. Why would you when, on graduation from a course much more difficult than all the people churning through law and finance, you can look forward to never earning more than a tube driver, and watching the continuing decline of British industry and your future employment prospects, all from the comfort of your overpriced hovel in Surrey.

If you come here you go jump on whatever fancy bandwagon is the latest trend (seems to be JavaScript), learn a whole bunch of buzzwords so you can out-bluff the guy who will be bluffing you about his knowledge in the interview, and then ride the contracting gravy train, churning out the same Angular implementations over and over again.

Again I tell you,

[Anonymous Coward]

it is easier for a camel to go through the eye of a needle than for a rich man to enter the kingdom of God - jesus

Good

[Anonymous Coward]

Now make it work reliably and cheaply over distances in the order of hundred, nay thousands of kilometers, and I am sold. Till then, go back to your labs and suppress the hype.

impressive, but...

[yodleboy]

I have a feeling it could be improved with systemd.

1.125 Tb/s over 121.5 GHz bandwidth?

[manu0601]

1.125 Tb/s over 121.5 GHz bandwidth? I remembered there was an information theorem that demonstrated this to be impossible, but I must be remembering wrong. What is the actual relationship between maximum throughput and analog bandwidth?