Catch up on stories from the past week (and beyond) at the Slashdot story archive

 


Forgot your password?
Close
typodupeerror
×
Science Technology

Hollow Optical Fibres Can Now Process Signals 108

Posted by timothy from the computer-is-the-network dept.
Ami_Chan writes: "According to Nature, researchers at Bell Labs have created a new type of optical fibre. This fibre is hollow, and can be tuned to different wavelengths of light using 'plugs of fluid' and temperature changes within the fibre. This allows the fibres to process signals as well as transmit them. The full article is here."
This discussion has been archived. No new comments can be posted.

Hollow Optical Fibres Can Now Process Signals More

Hollow Optical Fibres Can Now Process Signals

Comments Filter:

  • Here's the full text (Score:2, Informative)

    by qurob ( 543434 )
    [tt]Optical fibres currently do the boring legwork in telecommunications. Soon these light-filled strands may play a more active role. Researchers at Bell Laboratories in Murray Hill, New Jersey, have created optical fibres that can be switched between different states that transmit light of different colours1. These fibres can process signals as well as carry them.

    Devised by John Rogers and his colleagues, the new fibres are hollow. Perforated with channels thousandths of a millimetre across, each fibre looks like a bundle of drinking straws. Their tunable behaviour comes from plugs of fluid within that can be pumped back and forth.

    These 'microfluidic fibres' combine the cheapness and robustness of conventional fibre optics with the functionality of more complex and expensive devices. Currently, when switches or transistors are installed midway along the length of a fibre, they can end up buried and inaccessible along underground or seafloor transmission lines. Breakdowns in such cases are understandably costly.

    Wavelength-division multiplexing, for instance, is a common way of sending many optical signals down a single fibre simultaneously. Different signals, encoded in light beams of different colours, are unravelled at the receiving end using special filters or light sensors.

    Microfluidic fibres could act as both transmission channel and filter, and could be switched to relay first one signal and then another - without all the separate paraphernalia that is otherwise needed to decode the signals.

    The fluid plugs alter the fibres' light-conducting behaviour. Light travelling through the fibres' solid glass core changes when it passes through a region surrounded by fluid. Under certain conditions, this can make the fibre relatively opaque to light of a narrow band of wavelengths, so that the fibre filters it out.

    The filtered wavelength can be tuned by altering the temperature of the fluid; this is done by a tiny electrical 'heater' wrapped like a sleeve around a short section of the fibre. The wavelength and attenuation of the filtering can be controlled using a second heater further down the fibre, to warm up the air in the channels. This pumps the liquid plugs further inside or outside the region where they become active as filters.

    Rogers and colleagues anticipate that other arrangements of fluid plugs, heaters, pumps and so on will fulfil a variety of other functions that are needed in optical-fibre communication networks.

    References
    Mach, P. et al. Tunable microfluidic optical fiber. Applied Physics Letters, 80, 4294 - 4296, (2002).
    [/tt]

  • Wow! (Score:2, Funny)

    by Limburgher ( 523006 )
    I wonder if one of those fiber-optic lamps you can get would function as a CPU. . .

    • Re:Wow! (Score:3, Interesting)

      by digitalunity ( 19107 )
      I can't wait.

      The future I see coming out well before optical microprocessors:

      Field Programmable Fiber Arrays. They will be hybrid chips with semiconductor controls and fiber optic IO. The telecoms are gonna shit their pants when this stuff comes out. These are going to be ultra-high speed stateless DSP's, capable of outprocessing their electronic counterparts in magnitudes of superiority.

      Just imagine the benefits:
      Less latency
      Higher bandwidth
      On-The-Fly topology reconfiguration
      Learning switch fabric

  • AMAZING! (Score:4, Funny)

    by EvilAlien ( 133134 ) on Tuesday June 11, 2002 @02:01PM (#3681092) Journal
    This has an amazing amount of promise. The implications of this technology are staggering.

    Now I just wish I wasn't all wacked out on a coke slurpee and sluggish from lunch so that I could think about the implications and actually say something intelligent.

    • "This has an amazing amount of promise. The implications of this technology are staggering."

      I expect you could build some seriously uber-ninja neural networks with this stuff! This brings in applications from practically every genre of computing.

  • but? (Score:5, Interesting)

    by paradesign ( 561561 ) on Tuesday June 11, 2002 @02:02PM (#3681101) Homepage
    wouldnt this system be susceptable to distortion through vibration? if the line is bumped it would cause a shift in the fluid, if only a minute one, possiply distroying the signal. it would be interesting to se what measures bell labs has taken to account for this if amy at all.
    • It could be a simple matter that surface tension is strong enough to hold the plug in place. A droplet of water stuck in such a small place would be difficult to disturb.
      • but it could still be disturbed, and thats the problem. on the ocean floor anything could happen, seismic activity, volcano, depth charge? who knows.
        • Umm, do you think a regular ol' fibre line would be able to withstand seismic activity, a volcano, or a depth charge? Somehow I doubt it...
        • Who is really worried that someone will send depth charges down to disrupt international fiberoptic lines? (Al-Qaida, hint hint, you need an underwater cell next)
    • Yes it would be problem but you don't need to have this new fibre running the entire length of the cable across the ocean floor.Coupling this new fibre to standard fibre at the terminal end will be trivial - standard optical components are available for that - the new fibre and equipment can be kept in a controlled environment.

      Chang

  • A little short on technical details (Score:3, Interesting)

    by Christianfreak ( 100697 ) on Tuesday June 11, 2002 @02:06PM (#3681125) Homepage Journal
    I'd be interested to know exactly how it works. The article talks in length about heat and fluid changing the light but either I'm missing something or it doesn't really say how.

    Is it really changing the light or are they creating some kind of filter?
    • Sounds a lot more like an overglorified mechanical filter than something that will be used to create powerful digital devices.

      I guess I fail to see how this is better than a digital filter.
    • A technical paper is available via linkage at the bottom of the story...

      Mach, P. et al. Tunable microfluidic optical fiber. Applied Physics Letters, 80, 4294 - 4296, (2002). [aip.org] You might need a subscription to read it though, not sure.

      Basically the application is a wavelength selective filter for WDM. They are likely looking at arrays of these for WDM switching, as a fiber based alternative to MEMS.

    • The fiber has a central channel (core) that transmits the light by total internal reflection. This channel is surrounded by six large channels that contain the fluid/air mixture. The dimensions are such that the light transmission along the core is normally not affected by the presence of the fluid in the outer channels, except for a limited region where they have written a "long period grating" into the core. When the region with the long period grating has the outer channels filled with the fluid the corresponding change in the refractive index leads to a reduced internal reflection in the core and thus light leaves the core and is therefore "filtered out". The filtered wavelength depends on the refractive index of the fluid which can be tuned by the fluid temperature (the authors report 0.10 nm/K shift of the filtered wavelength with temperature).

      How is the fluid brought to the grating region?
      The outer channels are not filled completely with the fluid but only at a certain length - the rest of the channels before and after is filled with air. By heating the air portion of the fiber the pressure of the air increases and the fluid is pushed towards the colder air segment. Of course the channels have to be sealed hermetically - this is achieved by splicing the "active fiber" to a conventional single-mode fiber at both ends.
  • It's probably just another hoax [slashdot.org]
    </sarcasm>

    Actually sounds like switches might start keeping up with the bandwidth. Although keeping fluid and tubes at exact tempratures can't be cheap. Think superconductors.

  • This has significant ramifications (Score:4, Funny)

    by Anonymous Coward on Tuesday June 11, 2002 @02:07PM (#3681142)
    On the Lite-Brite product line.

  • Denial of Service... (Score:4, Funny)

    by ioexcptn ( 190408 ) on Tuesday June 11, 2002 @02:09PM (#3681157)
    ...with a heat gun. So all I have to do is heat the fibre up and completely destroy data integrity? Sounds like a great idea.
    • Yeah, well, if you melt regular cable you'll disrupt the signal too. Not much knowledge gained here except that I'm not letting you near my datacenter with anything flammable :)

      -Matt

      ---

      Got web hosting? RackNine Inc. [racknine.com]
    • Yeah, but doesn't a heat gun cost significantly more than a scissors? People can walk up to an ethernet cable and give it a snip to destroy data integrity. But you have a good point, because it would provide only temporary disruption.

    • with a heat gun

      ...which is much easier than with scissors anyway.

    • Actually having a fiber tha sensitive would be a wonderful idea for distributed sensing apparati. You'd have an extraordinarily sensitive, sheap, and disposable sensor for heat, compression, what ever you like - their is enormous possibility to fiber sensing. Boeing is now using it for Aircraft Development, and Norway has installed a fiber sensor system in a Boat for realtime, constant Stress Management.
  • Alright this does sound fascinating and all but fiber is already really expensive and so are filters for breaking down the transmission into it's composite "colors" but it seems to me that it wouldn't take much run of this new and presumably much more expensive fiber to pay for the filters.

  • Pump (Score:4, Insightful)

    by totallygeek ( 263191 ) <sellis@totallygeek.com> on Tuesday June 11, 2002 @02:15PM (#3681193) Homepage
    The story mentions fluid pumping. This means moving parts, which means more chance for failure. If the speed does not jump by several orders of magnitude, or the distance limitations disappear, I don't see why anyone would install the technology built around this.

    • You know, you can also "pump" fluid using electromagnetic fields...
      • You know, you can also "pump" fluid using electromagnetic fields...


        While true, now you have to provide power every so often along the "wire". Magnetic effects go less of a distance than photo-optic ones, so you are cutting down on your maximum distance without repeaters. This, too, is unacceptable.

    • The article mentions that you can use the heaters to expand and contract air gaps between plugs of fluid, which has the effect of moving the fluid around without the need for mechanical pumps.
    • in addition to the other suggestions on how to move the minute quanties of fluid we're talking about here, you can also use piezo elements to change pressure. just a thought.
    • Is it just me, or while reading this article I found myself thinking of a natural system (sortof like nerves and synapses)... it makes you wonder where all this is leading us...
  • Comment removed based on user account deletion

  • ...and (Score:3, Funny)

    by docbrown42 ( 535974 ) on Tuesday June 11, 2002 @02:20PM (#3681221) Homepage
    ...they make nifty drinking straws!


    -Ed
    docbrown.net [docbrown.net]

  • How (I think) it works (Score:4, Interesting)

    by lirkbald ( 119477 ) on Tuesday June 11, 2002 @02:28PM (#3681272)
    This sounds like it's based on the same technique used to make filters in the microwave band. By creating a transmission line with several appropriately spaced steps in the impedance, you can create a low-pass filter. With some more sophisticated branching of the line, you can make a high-pass or a band-pass filter. The technique relies on interference and reflection effects from the boundaries between the transmission line sections. I think they're doing the same sort of thing here; introducing fluid into the center of the line will change its refraction coefficient, which takes roughly the same role in the fiber that impedance does in a transmission line.

  • secure? (Score:3, Interesting)

    by GoatPigSheep ( 525460 ) on Tuesday June 11, 2002 @02:33PM (#3681312) Homepage Journal
    Would this type of wire be easier/harder to tap into than normal fiber optic wire?
    • Probably equally as hard. The way I understand it (as explained by an FBI guy), tapping fiber optics breaks the connection, literally.
      • It can be done. It cannot be done transparently. It takes a fair bit of time and equipment and even if you are sucessful, they will still be able to tell.
        • It can be done. It cannot be done transparently. It takes a fair bit of time and equipment and even if you are sucessful, they will still be able to tell.
          Actually every fibre optic cable suffers from leakage, it may be possible to detect the data in this leakage with a vulcan joint without breaking the connection. I don't remember the details.
          • it may be possible to detect the data in this leakage with a vulcan joint...

            Tapping via mind meld? on cannabis?
            • Tapping via mind meld?
              I'm just quoting an urban legend... (computing legend sorta)... Or maybe everybody that replies negatively to my post is actually a CIA agent trying to spread disinformation. I distinctly remember watching the Discovery channel and watching this US spy submarine (DSRV cover story) tapping into undersea cables by taking a feed from the repeaters.
              on cannabis?
              I follow God's law, not man's law.
        • They can tell if they're looking for it, at least. Any method of tapping a fiber optic cable would require taking some of the light out... thus weakening the signal... notice a signal strength drop and you know you're being tapped.
          • Right.

            And if you are dealing with a high tech DWDM transmission, you'd have problems tapping it without a repeater with a digital tap in it. This would cause latency, which would also be detectable.

  • "End-to-end" versus "smarts in the network" (Score:3, Insightful)

    by dpbsmith ( 263124 ) on Tuesday June 11, 2002 @02:53PM (#3681419) Homepage
    OK, I understand the researchers were doing very cool things that might have a whole range of interesting uses, but...

    I thought the whole lesson of the Internet was that the network should provide connectivity only, with a bare minimum of built-in processing...

    because, if you put processing into the network you are making fundamental assumptions about how the network is going to be used. In other words, processing within the network = optimizing for predetermined uses = locking out future evolution and outside innovation.

    Shades of the old Bell Labs that were committed to circuit-switching and opposed to packet-switching!
    • While there are certianly examples where this is true, there are counter examples. I don't think the Internet has decisively come down on either side of this debate. Look at web/streaming proxy/caches : there are numerous network topologies where these are clear wins. Admittedly, I wouldn't want a lot of "smarts" embedded in ASICs in my network, but the "network" is a lot more than routers and wire these days.

      And, on a completely different angle, I wouldn't be suprised if some/most of the uses of this are target ed towards non-network type of uses. For instances, using them for interconnects in a box or between two very localized boxes. Yeah, a network by some definition, but the "Internet" hasn't really taught us much about PCI buses.

    • The whole point is: the ideal network is an unlimited broadcast network. Every host can see every packet, and we have unlimited bandwidth.

      Another way to put it: bandwidth cures switching.

      That said, that's not possible; we don't have it. We have to have a way to optimize the links we use.

    • Not everything needs to go to service the Internet. Even if this is what this is being developed for, the physical signals to make the fluids do one thing or the other can be independent of what is going on traffic wise or can be the physical layer manifestation of whatever the upper layer decides.

      No?
    • This microfluidic optic fiber is not a network, it's only a short piece of fiber that is inserted in a conventional fiber. It's more like a switching device than a network. The control signals have to be electric, too (you have to operate the heaters), so the logic is not part of this fiber. The microfluidic fiber is only the device that does the actual filtering. The advantage of this fiber is that you don't have to seperate all the different wavelength modes before you can filter some of them out, but you can do it directly in a fiber.
  • As long as they're careful with it... don't want to open up any portals to "fluidic space;" we might get all sorts of horrible beasties coming through.
  • "In related news Bell Labs has made first contact with a new race of being said to live in the fluidic space created within a new breed of fibre optics..."

  • light wave rather than light flash (Score:3, Interesting)

    by ArsonSmith ( 13997 ) on Tuesday June 11, 2002 @03:04PM (#3681499) Journal
    will fiber optic ever be changed to an analog version with the wave of the light being the transport of data rather than the flashing?

    It seems that the data would be moved much faster if the sensors were able to pick up on individual light colors and waves rather than just on/off of the light. This would be able to work similar to how a modem works with diffrent tones producing diffrent characters, etc...

    ps.
    If this is already how fiber works than my understanding is just way off and please disregard.

    • Hi,

      Optical fibers don't work that way, at least at the high end.

      High end optical fiber equipment uses light on several different carrier frequencies (or wavelengths or "channels" or "colors"---it all means the same thing). This is not unlike your analog car radio or TV. However, each of these channels tends to be a digital data stream.

      For example, a state-of-the-art DWDM (dense wavelength division multiplexing) system could have a single optical fiber carrying 128 channels in the 1500nm to 1600nm band where each channel is separated by 50GHz. Each channel contains a stream modulated at somewhere in the range of 10-40Gbs. The modulation scheme for the channels tends to be some kind of digital scheme (NRZ and RZ are two common methods).

      To put things in perspective, the net bandwidth of current (but not necessarily deployed) optical fiber equipment:

      128ch * 40 Gb/s ~ 5 Tb/s

      Research results at faster speed have been demonstrated but nobody is buying equipment right now. (See the whole dark fiber problem ... the technology has outpaced the demand.)

      Kevin

      (P.S. I have a Ph.D. and I work in the same building.)

  • Background (Score:2, Informative)

    by photonic ( 584757 )
    This technique is based on photonic bandgap fibers. They were invented by Phillip Russell at Bath University [bath.ac.uk] (UK). These fibers contains a pattern of hollow channels that form the 2D equivalent of a multilayer mirror. Light can not travel in such a region. One channel in the middle of the matrix is missing, creating a defect state where the light can travel.
  • Discover Magazine [discover.com] also has a blurb about this research in the R&D section of their June 2002 issue. Or, read it online instead [discover.com].
  • How about a CPU that has one fiber line for each opcode, and the processing happens as the signal passes down the pipe: 1 "cycle" per op. New parallel arch? Can't bandgaps do this?

  • This seems like it could circumvent some of the issues currently preventing wiretapping of optical fiber... That's not really a good thing, now is it? =D
    • Relying on the physical transport mechnanism to keep data secure is not a good idea. People should assume that what they are sending can be captured by someone you don't want seeing it. I think a strong point to point encryption scheme is the better way to go.
  • ...FIBER!!
  • Hrm won't this kind of be like the AI computers in the hyperion series which inhabit the medium of communication?
  • I wonder if this could be used for error detection on long links. Like maybe have the checksums verified every so often. If the wire itself could drop a corrupt datagram it would save the devices on the endpoints some effort. The upper layer protocols wouldn't care, all they know is that a datagram was lost which they can handle.
    • This fiber device does not look at the data at Gb rate - it's just a switch for different colors of light (at a millisecond rate, way too slow for any ECC logic). Also, the logic of when to switch what is not part of this device - that logic has to be brought to the device as electrical signals from outside.
  • It used to be that Bell Labs would announce something and I would take it at face value. But lately they've had some trouble with verification of claims. Does anyone know if there has been any outside verification of this work?
  • Connection bogged down? Throughput is kaput? Dump in the Draino! or call a plumber!

Slashdot Top Deals

My sister opened a computer store in Hawaii. She sells C shells down by the seashore.

Close