back to article Record-breaking Aussie boffins send 44.2 terabits a second screaming down 75km of fiber from single chip

Australian scientists say they have broken data communications speed records by shifting 44.2 terabits per second over 75km of glass fiber from a single optical chip. The five-by-nine millimetre prototype gizmo is described as a micro-comb in a paper detailing its workings, published in Nature Communications on Friday. Light …

  1. cbars

    My hat is off!

    1. abs

      I think I know what my next lock down movie will be...

      Dark Helmet: "What have I done? My brains are going into my feet!"

    2. Anonymous Coward
      Anonymous Coward

      For what?

      Kudos to the academics but but down under pretty much 90% of the population is stuck with 25-90Mbps.

      There are some states that have gone the gigabit but it is nonetheless a distant dream for almost everyone else.

      1. Slipoch

        Re: For what?

        If you were lucky enough to get the original FTTP/FTTH you can do around 4Gbps on current plans, capable of so much more currently, tack this on at either end and you have such fast speeds you end up in enterprise areas much easier and cut out the middle-men.

        Besides Australia makes a ton of money each year off our patents with Wifi, this'll be something else the govenrment can fire the inventors of then use the profits to help rugby.

  2. Kernel

    DWDM on a chip

    Very impressive!

    The only problem I can see is that quite a large number of racks are going to be needed to MUX/DEMUX enough lower order trib streams to produce a single 44Tb payload for the single chip DWDM.

    1. Christian Berger

      Actually it's even more impressive

      It's not just DWDM on a chip, it's OFDM on a chip. The carriers can be much closer together relative to their bandwidth than on DWDM as they are all coherent and have a well defined distance. Essentially the neighbouring frequencies will interfere with the middle one, but those interference patterns will cancel themselves out over a symbol period.

      Honestly I do not think this will require a lot of rack space. After all this is "just" 422 times as much as already established 100 Gbit/s and this can apparently is compatible with CMOS, so you could even place your routing logic on the same piece of silicon.

      1. MiguelC Silver badge

        Re: Actually it's even more impressive

        “We used a next-generation optical modulation format with 500 gigabits-per-second per wavelength. With the 80 micro-comb wavelengths, this combined to form an optical super-channel, which added up to 40 terabits per second,” Corcoran said.

        Honest question: how can they achieve 44Tbps with 80 500Mbps wavelengths? Arithmetic didn't change, AFAIK...

        1. Solviva

          Re: Actually it's even more impressive

          44 Tbps includes protocol overheads. 80 times 500 Gbps (not Mbps as you wrote) is a nice round 40 Tbps, so add on 10% for overheads and you get an actual rate of 44 Tbps.

    2. Fred Flintstone Gold badge

      Re: DWDM on a chip

      I just want to see where they'll fit this on a shark, being this small..

      1. Anonymous Coward Silver badge

        Re: DWDM on a chip

        Left nostril

    3. gerryg

      Re: DWDM on a chip

      You're missing the point. The fibre they're using is in the ground already. The cost of the system was never the cost of the fibre nor electronics, it was the cost of digging the trench. Everyone, forever, knew the potential bandwidth of single mode fibre but lasers, in 198x, were a bit rubbish and detectors were not amazing either. But back to the trench, stick a dozen fibres in the ground even though you don't need eleven of them, because you don't need to dig a second trench and leave dark fibre there because you didn't know how fast Tx/Rx was going to improve. Research on Wavelength Division Multiplexing was also going on in 198x, but in to give a comparison in those days line width on VLSI was in 10s of micrometres.

  3. Mike Shepherd


    "...due to the unprecedented number of people using the internet for remote work, socialising and streaming porn".

    Plus ça change...

    1. Phil O'Sophical Silver badge

      Re: Correction

      Maybe not. There was a story in the papers a week or so ago about a sex therapist who was finding an increasing number of people who were unable to get enough privacy in lockdown for their usual solitary relaxation.

    2. Anonymous Coward
      Anonymous Coward

      Re: Correction

      But porn only takes about 5 minutes per session, so can't be that demanding.

      1. Evil Auditor Silver badge

        Re: Correction

        You're young, aren't you?

        1. Old69

          Re: Correction

          "You're young, aren't you?"

          Note "per session". For the young the repeat interval can apparently be very short. Not sure what is claimed as a record. IIRC The Kinsey Report did have some double-digit*** numbers per day for what was considered the male capability peak circa age 16.

          ***not necessarily double-finger

    3. Rol

      Re: Correction

      The University of SMUT is currently working on a porn AI, and they are optimistic that their program will be able to generate any and all customisable flesh combinations with associated grunts and groans to fool anyone into thinking it is real.

      This program, once installed on a PC, will free up 90% of the internet and reduce the instances of PC infection to almost zero.

      1. stiine Silver badge

        Re: Correction re: porn AI

        Let me know when there's a Zoom api module.

      2. batfink

        Re: Correction

        Well, for some forms of infections...

  4. Anonymous Coward
    Anonymous Coward

    Any idea when it's coming out?

    Thinking of upgrading my subscription to my provider of romantic films.

  5. redpawn

    It's the coin

    Inspired by the article I epoxied a coin to my cable modem and my throughput went up by more than an order of magnitude. Another doubling came from wrapping the cable line around a magnet.

  6. Anonymous Coward
    Anonymous Coward


    And here I was thinking all they could come up with Down Under was the rotary clothes line.

    1. The Central Scrutinizer

      Re: Cool.

      Yeah, not much, just things like wi-fi, black box flight recorder, spray on skin, bionic ear.... etc.

      1. eldakka

        Re: Cool.

        ... Salt-water crocs, stingers, irukandji, funelweb, taipan, death adder, dog-drowning kangaroos, peacock spiders ...

  7. TheSkunkyMonk

    Actually it is ideal for home broadband now! as long as the cabling is good probably wont be any need to replace the system for the next 100yr or so. Mind we hate that type of thinking in government projects, why make something to last when we can have lucrative maintenance contracts to put out to tender each year or a system we can upgrade every few year.

    1. anonymous boring coward Silver badge

      The last bit to the home is copper. Old copper from decades ago.

      It's already been pushed to 70Mbit or so, which is amazing when you think about it.

      But having 800 carriers in an optical cable won't alter anything on the copper end.

      1. Martin-73 Silver badge

        No fibre to the [premises/home/business/manshed] in .au?

        1. sms123

          I have FTTH it just depends on where you live.

        2. anonymous boring coward Silver badge

          No fibre to the home in UK, in the vast majority of places. Investment in infrastructure is not a big thing in the UK. We have Tories in charge here.

    2. Pete4000uk

      Apart from that, WHAT have the Australians done for us?

  8. IanDs

    The issue isn't so much getting enough data in -- there are already switch chips with 25Tb/s capacity, with 50Tb/s coming in the near future -- as what happens to the optical signals. Optical transport networks rarely treat the whole fiber bandwidth (>40THz) as one point-to-point pipe (one massive "superchannel"), they use ROADMs (Reconfigurable Optical Add-Drop Multiplexers) to route different wavelengths to different destinations. The closely-packed signals of the "optical comb" approach can't be used for this because you need guard bands between the channels to allow them to be separated and recombined -- and also they don't all come from the same source.

    There are parts of the OTN system where one fat pipe crammed with data is all that is needed (e.g. undersea links of several thousand km) but these are only a tiny fraction of the total installed hardware, far more is used for shorter links within countries and continents and these all use ROADMs.

    There are also issues to do with yield and reliability when you integrate so many optical channels into a single device.

    So it is an impressive technical achievement, but don't think it's going to completely transform the world ;-)

    1. tip pc Silver badge

      “ There are parts of the OTN system where one fat pipe crammed with data is all that is needed (e.g. undersea links of several thousand km)”

      RFOG is often used in long distance transmission as the signal can be amplified several times before it needs to be recreated.

      This new technology is not for long haul transmission. Its core, site to site & transmission And will enable faster edge.

      1. Jellied Eel Silver badge

        This new technology is not for long haul transmission. Its core, site to site & transmission And will enable faster edge.

        I guess this is where pricing will come in. On shorter distances, infrastructure owners may already have 144f of 288f cables and the ability to run more in their ducts. So potentially cheaper to just light another fibre. And possibly less risky, ie 44Tbps of credits vs 25Tbps in an outage. I guess it could be handy for data centre replication/backups where DC's are 50-75km apart, or of course for academic networks where they're generating collosal amounts of data, ie the benefit if it being wider, not 'faster'.

  9. Barrie Shepherd

    Staggering - I was impressed when I started work in 1970 and we were squeezing 960 telephone channels down a 4 MHz coaxial cable! Then we got 30 channels down an ordinary telephone quad and the rapid expansion of digital comms began.

    This development just shows how innovation can continually improve the things we thought were already at "maximum smoke"

    1. Kernel

      "Staggering - I was impressed when I started work in 1970 and we were squeezing 960 telephone channels down a 4 MHz coaxial cable! "

      Aah - you're of the same era in the telecommunications business as myself (I started in 1972, retired October last year).

      The good old days, when a supergroup was massive capacity and digital communications meant a VF telegraph system running at 30 baud.

  10. John Brown (no body) Silver badge

    a micro-comb in a paper

    Is this to accompany the tiny violin?


    Only part of the problem

    75Km is not very far. So you have crack regenerating the signal as an electrical signal.

    Would it not be easier to work in packing more fibres into the same physical cable. Better sending signalling over longer distances in one hop.

    1. This post has been deleted by its author

    2. AndrueC Silver badge

      Re: Only part of the problem

      75Km is not very far.

      That depends what and where your target application is. In the UK that would probably be enough for local backhaul. Although (at the risk of being quoted forevermore) I struggle to believe we'd ever need that much bandwidth for that.

      1. Anonymous Coward
        Anonymous Coward

        Re: Only part of the problem

        "at the risk of being quoted forevermore"

        It's a given for any statement of an absolute limit. Even quoting "The Laws of Physics" when referring to latency might not hold true if quantum effects are able to be exploited to exceed the transmission speed of light.

        There is a quote along the lines:

        "When an expert says something can be done - then they are usually right. When an expert says something cannot be done - they are likely to be proved wrong eventually."

        Parkinson's Law is immutable: "Everything always expands to use all the available resource".

    3. Jou (Mxyzptlk) Silver badge

      Re: Only part of the problem

      Oh, but it is far.

      They will remodulate every 40 km, usually a lot less. Even the sea cables remodulate at less than 40 km (I remember each 10 km, but I'm not sure).

      Check that table on German wikipedia

      The English wikipedia article is, sadly, not as detailed as the German, compare if you want .

    4. Kernel

      Re: Only part of the problem

      "75Km is not very far. So you have crack regenerating the signal as an electrical signal."

      No, regeneration only happens at 100's or 1000's of km intervals, determined by the capabilities of your specific fibres, error correction capabilities of the coding scheme in use, etc.

      Analogue amplification of the composite optical signal is a different story though - yes, that does happen every 70km or so.

    5. Will Godfrey Silver badge

      Re: Only part of the problem

      They only said they succeeded in doing that, not that they couldn't have gone further. It might be the longest piece of fibre that had available

      1. coconuthead

        Re: Only part of the problem

        It is suspiciously close to exactly 4 times the distances between the two campuses of Monash they said they tested between. (It is 18km from Clayton to the city.)

    6. eldakka

      Re: Only part of the problem

      75Km is not very far.

      Hate to break it to you, but undersea cables currently have repeaters/amplifiers every 70-150km on them already. It's not a single unrepeatered/unamplified run of 5000km (or whatever the actual distance is, I don't have a ruler that long) across the Atlantic.

      75km puts it within the range of current, unamplified distances (at the lower end sure, but still within the range).

  12. Anonymous Coward
    Anonymous Coward


    With all due to respect, we are stuck with 50Mbps and some places even worse and we are talking about Terabits?

    Is this a joke?

    1. AndrueC Silver badge

      Re: Disgusted

      There's more to computer networking than the final mile, you know. Something has to connect head-ends/exchanges together. Something has to carry signals between ISPs. Something has to carry data between countries. Those somethings have to be high capacity because they carry the data being consumed by end-users such as yourself.

      If every residential property in the UK had 50Mb/s (which on average it probably does) that's over 140Tb/s. Now as it happens no-one gets or needs 1:1 contention so there's a lot of sharing of bandwidth going on. But still, cables such as these are going to be very useful over the next few years.

      It's unlikely they will ever be used to connect to your property but they could end up being used deeper in the network.

      1. anonymous boring coward Silver badge

        Re: Disgusted

        "If every residential property in the UK had 50Mb/s (which on average it probably does) that's over 140Tb/s."

        By a factor of about 10, I reckon. Unless my brain failed me, which it admittedly frequently does.

    2. Jou (Mxyzptlk) Silver badge

      Re: Disgusted

      No, it is bad reality to leave large parts of the country in the communication stone age of less than 6000...

    3. Stuart Castle Silver badge

      Re: Disgusted

      And what do you think carries the internet data to and from your ISP's network, which is likely carrying tens of thousands of connections that are anywhere from 1Mbps to 1Gbps?

    4. Mike 137 Silver badge

      "stuck with 50Mbps ..."

      <bogus Yorkshire accent>"You're in clover. We get half a megabit at best..."</bogus Yorkshire accent>

    5. Anonymous Coward
      Anonymous Coward

      Re: Disgusted

      Nah, it just ensures you will have positively no problem with back end bandwidth contention.


  13. Wayland

    Practical test

    So they sent test patterns obviously so they could check it for errors.

    It would be nice to leave it in situ hooked into a normal router so it actually does useful work.

    The headline speed is impressive but simply making it compatible with existing gear will bring it to market.

    1. TRT Silver badge

      Re: Practical test

      One wonders if long-term use causes enough heating to stress the comb and shift the frequency spreading. Although I guess with that kind of device it has to have rock solid thermal management - i.e. a massive heatsink, like a lump of silver, and a Peltier to bleed off the excess heat in a controlled way.

  14. Bogbody

    Openreach 5 years to 40T

    5 years? .... theres a couple of nougts missing there ... heres a couple for you .... "0" "0"


  15. anonymous boring coward Silver badge

    "44.2 terabits a second"

    That's terafyingly fast!

    1. John Brown (no body) Silver badge

      Quick, call out the anti teranbit squad before something blows up!

    2. Kevin McMurtrie Silver badge

      What's terrifying is if the government tells you to log all packet headers for national security. Your one rack of comms would need 500 racks of spinning rust drives.

      You couldn't sell a 44Tbps link in America.

  16. The Aussie Paradox

    Another government Project

    Looks like we know what to do with our $60 billion surplus from the loan. We can phase out our ageing IPoAC** network.

    Those poor pigeons must be exhausted after the recent increase of usage since the lockdown.


  17. TeeCee Gold badge

    Missed a bit.

    ...dropping to 39.2 Tb/s and 39.0 Tb/s for the lab and field-trial transmission experiments...

    ...and about 20Mb/s when deployed by Virgin.

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Other stories you might like