'Speed of light' fibre optic breakthrough hints at faster internet

'Speed of light' fibre optic breakthrough hints at faster internet

Summary: Researchers at the University of Southampton have developed a hollow-core fibre optic cable they say has reduced the latency of data transmission.


Researchers at the University of Southampton claim to have increased the speed at which data can be transmitted over fibre optic cables, one of the links which form backbone of the internet.

Writing in Nature, the team said it has developed a hollow-core fibre that has transmitted wavelength division, multiplexed data at 99.7 percent the speed of light in a vacuum.

The faster light can travel through a fibre the lower the potential latency of data transmission. In comparison, the solid silica glass fibre optic cables in use today propagate light at some 69 percent the speed of light in a vacuum.

The difficulty with using hollow-core fibres to transmit data to date is that the cables have struggled to maintain the combination of low loss, wide bandwidth and mode-coupling characteristics needed for high capacity data transmission.

The researchers said they have now achieved an acceptable balance of these factors, recording a loss of 3.5dB per km, and using a 160nm bandwidth channel to transmit 37 WDM (wavelength division multiplexing) channels at 40Gbps, which they claim is 1.54 microseconds/km faster than over conventional fibre.

Apart from the general benefit in faster transmission, wide bandwidth, low latency signal transmission links are in demand to increase the rate at which algorithmic financial trading can take place and as fast interconnects between components inside future exaflop-scale supercomputers.

Topics: Networking, EU, United Kingdom


Nick Heath is chief reporter for TechRepublic UK. He writes about the technology that IT-decision makers need to know about, and the latest happenings in the European tech scene.

Kick off your day with ZDNet's daily email newsletter. It's the freshest tech news and opinion, served hot. Get it.


Log in or register to join the discussion
  • High Loss

    3.5 dB/km loss? That's over 10x the loss of solid silica glass.

    Interesting concept, but that loss would make it quite expensive to deploy over any distance - which is where the per km latency benefit would add up.
  • microseconds

    To me 1.54 microseconds/km is not much faster than before, a microsecond is a very short period of time. I suppose over a long distance it would help in the long run, but to the average user, it is not noticeable, it might make that European site load faster, even for East Coast to West Coast it would be only about 7.5 miliseconds better. (Using a 3000 mile distance)
    • microseconds = eternity

      As the article stated, we're not talking about the time that it takes to load a webpage. High speed computer algorithms trading market securities compete with each other on the scale of microseconds. Shaving a few milliseconds off the time from Chicago (CBOE) to New York (NYSE) puts you at a enormous advantage.
      • Another issue is how you handle the 6 times the equipment.

        If you optically amplify (light pumping) or using a media converter (in one SFP and out the next) after breaking the WDM into the separate lambdas, etc, then the transit time end to end will still remain faster than solid core. Light pumping however doesn't necessarily reshape the pulses into nice clean square waves.

        However if you convert to electrical, switch it (maybe to handle CRC errors, retime, etc) and then go back to optical (even once) you have just lost all gains from hollow fiber.

        If you can charge $3M per lambda per year vs $300K then the ROI of having a facility every 16 - 20 KM to regen is doable.

        This is a start but need longer hops.
  • Interesting concept BUT not ready for real world.

    Here are some specs today.

    For multimode fiber, the loss is about 3 dB per km for 850 nm sources, 1 dB per km for 1300 nm.

    So MMF is equal to or 3 times better than the loss for hollow core.

    For SMF the loss is about 0.5 dB per km for 1310 nm sources, 0.4 dB per km for 1550 nm.

    So SMF is 6 to 7 times better than the loss for hollow core.

    This means that on a 100 KM haul that would normally use a single SFP to drive the distance you now need 6 times the number of SFPs for 6 times the cost in electronics to drive the same distance. That means more buildings along the way with more power, heating and cooling to be included. More truck rolls, etc. 10GE DWDM SFPs are around $4-6K approximately but how much is the cost of each additional building along the way?

    So is 154 microseconds saved time, over 100 KM, worth $1M in increased cost for electronics and facilities? Fiber providers will have to figure this out but its not rocket science.
    • Depends

      Spread networks ran a more direct line from Chicago to New York and essentially put the other high speed links out of business. We're talking about algorithmic trading of financial securities. If you are a few milliseconds behind the competition then you are out of business.

      "As the path of lowest latency, Spread could command rates dwarfing those of other lines. Allied Fiber, which runs dark fiber through the longer railroad-plotted paths, asks $ 300,000 a year for its service. Multiple sources who negotiated with Spread said they were asking for eight to ten times the going rates, as much as $ 3 million per year."

      Steiner, Christopher (2012-08-30). Automate This: How Algorithms Came to Rule Our World
    • With respect...

      With respect, I disagree. I think there are very much real world applications for this. Not necessarily on a grand scale, but for intra-house networking, the efficiency gains would make this a very attractive proposition.

      On a commercial level, consider a server farm. These cables connecting each, giving a 30% efficiency gain across the servers. Not insignificant.

      On a domestic level, consider that we're mostly using Cat-6 cables at the moment. Good enough, but still limited by the copper in them. At some point, that copper wont be able to handle the potential speeds we'll be enjoying. A simple rollout of 10-15m hollow fibre lines instead for home networking is a logical step forwards. Specifically for situations where wireless isnt effective.

      There ARE real world uses for it, even if they are several years away. I expect that by the time this is ready for the real world, we'll be thankful they are there.

      Dont think of this as a global solution, I think you're correct that its not practical. Think of it as a personal solution, transfering data around a building, or even a suburb if you want to get expansive. The data loss at that point would be undetectable. The first step of the tracert, not the whole chain.

      Find some other solution for the backbone, but this has a chance at the ends.
  • latency IS a big deal

    I can attest to the importance of latency. I trade currency pairs on the Forex market. There are any number of 'automatic' trading algorithms you can use and/or create your own. Most of them are not profitable if you run them from your home or office due to the 10 millisecond to 100 ms latency between seeing a bid or ask price and your return order arriving at the trading desk. If on the other hand you run the program on a virtual server at the trader's co-location where the latency is on the microsecond level or less, the software is profitable.

    The big dogs push orders of millions of dollars back and forth all day every trading day and they can make a large profit on a few cents a trade edge, but only if they beat the OTHER guy to the punch. Since many of them are using similar trading strategies timing is everything.
  • gotta love this stuff

    " ... The faster light can travel through a fibre the lower the potential latency of data transmission. "

    No freaking sh|t Einstein.

    That's like saying "the faster a car travels, the higher the speed its moving at."

    Other than that Mr Stupidly Obvious (aka Nick Heath) .. go USH (University of Southampton).
  • Q

    Please advise I would like to write on plus500 review system its in my site www.ourplanet.co.il how I can publish in ZDNET also. Thx
    נועה בן פורת