Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 


Forgot your password?
Close
typodupeerror
×
Security Science

Holographic Sonar Cryptography 182

Posted by timothy from the bouncy-bouncy dept.
Atomic Snarl writes: "New Scientist.com has this story on how to encrypt a underwater sonar message using multiple sound path timing. By detecting and adapting for the current variations on underwater sound channels, the transmitted message can be received intelligibly only at a single point. This holographic approach suggests a method of web encryption using multiple hop paths and ping times to create a message which can only be decoded when received at a specific target node!"
This discussion has been archived. No new comments can be posted.

Holographic Sonar Cryptography More

Holographic Sonar Cryptography

Comments Filter:

  • Secrecy by Delocalization (Score:3, Interesting)

    by Spootnik ( 518145 ) on Wednesday October 24, 2001 @03:48AM (#2471016)
    Exactly, they aimed more at reliability -- even though the codes were lossy and reliability was achieved mainly by coherent en/decoding, because noise is incoherent. However, much of that was dropped in favor of faster and better (in that approach) use of homomorhic processing and other DSP techniques.

    Further, Holocomm's "delocalization" feature can be seen also in SHA-1, where *all* output bits change when one changes a *single* input bit. However, SHA-1 hopelessly mixes and merges all the data (as it is intended to do), while Holocomm allows for reversible and selective delocalization.

    Thus, in two contrast points to former pure holographic codes, Holocomm aims at (1) non-lossy reversible (2) selective delocalization -- which also allows interoperation with all known cryptography algorithms (that require exact data for decoding). The reliability feature is also further enhanced by the non-lossy aspect of it. As mentioned, Holocomm can also work in lossy modes, including lossy compression -- which can be quite useful.

    Holocomm is the first example of a practical quantum mechanical communication and encoding system that affords privacy and reliability, to a high degree, while also offering compression and selective information delocalization.

    As such, it naturally has many parallels in several things that are based on wave functions or on the Schroedinger equation .. which essentially defines wave phenomena ... as the theoretical basis of Holocomm, as stated.
    • Further, Holocomm's "delocalization" feature can be seen also in SHA-1, where *all* output bits change when one changes a *single* input bit.


      Wouldn't that mean that if you changed two input bits, all output bits would stay the same?

      • With this construct, if one input word is changed, the probability of a nonzero difference is ((2**N-1)/(2**N))**L (where L is the number of FFT layers, and the log2 of the number of input words), and furthermore, output differences of 0 are correlated in an easily detectible way.
      • if taken litteraly, then yes. An example of such a function would be parity.

        however, to be precise, SHA has every output bit influenced by every input bit. As does every other block cipher that comes to mind.
    • For years, I've been reading about the idea of data transmission using quantum entangled pairs of particles (possibly photons). The idea (Bell's Hypothesis) being that measurement of a property (eg spin) of a quantum particle will affect the property of another particle (which it has previously interracted with) instantly. That's instantly - not at the speed in light. This has been tested in the lab and proved to be true.

      This effect could be used for communication and would imply two things:
      1. As stated above, the communication would be instant, regardless of distance.
      2. It is impossible to intercept the message with affecting it as any measurement will affect the result.

      If it could be made to work, then you would have instant, uninterceptable communications. The problem being how you separate entangled pairs and get them to each end of the line. It's only been tested with distances of about 10 feet so far.
      • This is a really interesting technology, but I don't think it will be useful for us commoners quite yet. And I don't think I would have one either - for once, I believe the speed of this thing will be much lower than normal wires for some time yet, and it will be very expensive.

        But the possibilities are countless - Imagine how much easier it would be to control a space probe on Mars with zero latency!
        If this thing really works over such great distances this could be one major step ahead for space colonization and long-distance communication.

        (About encryption - it might seem like a swell idea, but remember that the particles have to interact some time before separating them, and then it would be just as practical giving your trusted party a symmetric cipher key instead of a molecule.)
      • I have my doubts about the use of entanglement for long (I mean really long!) distance communication for one simple reason - Although the entanglement effect is instantanious, the particles involved in the entanglement still have to travel from source to destination - and hence are restricted by the speed of light. Of course, for really long distances (interplanetary or perhaps even interstellar) some system of generating the particle streams ahead of time, and capturing the entangled particles at the destination in some sort of buffer at the destination could be envisioned but that would mean that (a) bidirectional streams would have to be generated for every source/destination pair of nodes and (b) sufficient particles for future use would have to be generated. Once a set of particles has been used to read/send a message, the entanglement is lost and a fresh set of particles would have to be used for the next message - this may well limit the usefulness of such a system. so, if you wanted to send a message a few light minutes/hours, this may not be a big problem, but if you were using it for interstellar traffic, you would have to estimate bandwith needs years in advance.

        • Yes, you are correct I think
          But it's even worse, because you won't even
          solve the latency problem (not easily at least)
          Because, even with thios scheme where you already
          send "preloaded" Exabits of data for future use
          when you instantly change their value,
          you will still need to let them know you did so.
          but if they measure somthing before you sent the message,
          then the roles are inverted :)
          So you would need (maybe) another load of boggus
          data, whose purpose would be to be constantly chewcked
          by your remote fellows to see appear "self obvious" messages
          like "ok, we did it now, check the data !"
          But, as you wish 0 latency, you'll have to check
          the bogus triggers so often, that you'll have an interesting probability to have self-obvious
          messages appear randomly now and then
          (remember you can't have initialized data, their value is unknown at "entanglement creation"
          It only become previsible at measuring time
          and only by the one that is measuring against his own
          data-stream, giving it the value of the actual data-stream.
    • Further, Holocomm's "delocalization" feature can be seen also in SHA-1, where *all* output bits change when one changes a *single* input bit.

      <NITPICK>

      Due to the nature of bits (being 0 or 1), changing a bit means flipping them from 0 to 1 of vice versa. Changing *all* bits, would mean flipping them all, i.e. a XOR operation.
      Changing a single input bit will change *some* output bits, not all of them. Would be a pretty useless hash algorithm ;-)

      </NITPICK>
      • Changing *all* bits, would mean flipping them all, i.e. a XOR operation
        <NITPICK>

        And to think that all these years I've been under the assumption that bit flipping can be accomplished with a NOT ;-)

        </NITPICK>

    • That's some mighty fine technobabble you've assembled there. If you're not already there, I strongly recommend heading out west and becoming a Hollywood script writer. They need people like you, to make the characters seem smart without actually saying anything.

      Seriously though, I should point out that in a hydrological holographic communications medium, the thermal dynamism will, while not affecting the message in, a, lossy way, will cause changes in the order, that, the binary components might be received. It is of course trivial to correct for this, current use of super string theory provides an elegant method for accounting for said brownian reverberations in the stream, but it's still important to XOR the bits into a checksum before sending, just to make sure you understand.

  • Speed of sound versus ping times (Score:5, Insightful)

    by Chiasmus_ ( 171285 ) on Wednesday October 24, 2001 @03:49AM (#2471017) Journal
    It seems to me that the speed of sonar through water is a physical certainty; that's why we can accurately use it to detect the distance from an object.

    Internet traffic is another matter. If I tried to use a ping time to measure the geographic distance to another server, I'd be about as scientific as the Slashdot poll.

    Am I wrong, or could internet latency give or take 100 ms or so from a ping, rendering the encrypted message readable by.. no one?
    • I think the speed of sound in the water has small variations due to pressure & temperature shifts. And error checking & correcting would be difficult, as these variations would make the message unhearable.

      To me, it seems as hazardous underwater as on the Internet.
      --
      • I will let you calculate it. The following are empirical equations from Kinsler and Frey.
        t is in degrees C and c is in m/sec.

        In Fresh water c = 1403 + 5t - 0.06t^2 + 0.0003t^3
        Good for 0 to 60 degrees C.

        In sea water
        c = 1449 +4.6t - 0.055t^2 + 0.0003t^3 + (1.39 - 0.012t)(S - 35) + 0.017d

        Where S is the salinity expressed in parts per thousand, and d is the depth below the surface in meters.

      • Re:Speed of sound versus ping times (Score:5, Interesting)

        by Ronin Developer ( 67677 ) on Wednesday October 24, 2001 @09:12AM (#2471664)
        The sonar conditions vary considerably through time. There are inversion layers and tunnels that are formed due to the differences in the index of refraction for the audio signals.

        In optical holography, you are recording the interference patterns resulting from a reference beam and reflected light. When you shine a laser of the right wavelength through or off the hologram, the interference patterns are "replayed" thus reproducing the image.

        Little if any information can be gleamed from a single intererence pattern.

        In the case of sonar, you are recording audio interference patterns. However, unlike in an optical holographic environment, the conditions change drastically under water depending upon weather condition and seasonal (or even geophysical (i.e earthquakes and volcanos) variations.

        In a controlled scenario as described in the article, it works because the replay occurs in a very short time period and the interference patterns may not change much. Without an initial reference signal, it may be very hard to get a good mapping of the sonar environment.

        As for the security, I wonder if you recorded the signal eminating from a single transducer at short range if you could actually receive the message at a spot other than intended.

        RD

    • Asymmetric routing makes this moot anyway (Score:5, Interesting)

      by Kenneth Stephen ( 1950 ) on Wednesday October 24, 2001 @04:16AM (#2471083) Journal

      Even if you could eliminate the problems with the latency, the asymmetric routing that exists in the internet will kill this technique. This communication technique depends on the forward and the reverse path being identical - something which is not true when asymmetric routing is used.

    • If the encryption were actually based on ping times, you're right, this would definately not work. But, it doesn't have to be. It does bring up a fairly good idea that, to my knowledge, has not been successfully implemented.

      Why does an encrypted message have to be sent all at the same time, and therefore by the same route? If you were to split a message into an arbitrary number of pieces, with each one getting to its destination through a slightly different route, it might be a little more difficult to intercept.

  • ...which leaves the question...

    Does this mean that they need more "big rocks" under the Great Lakes, or can they still use the same "big rock" to use this?

  • Radio? (Score:3, Interesting)

    by redcliffe ( 466773 ) on Wednesday October 24, 2001 @03:49AM (#2471020) Homepage Journal
    Could this be used to secure wireless networking? This would be an ideal way, because it is only understandable at one location. I don't know if it would work well though on Seattle Wireless or Brismesh style 802.11 networks.

    David

    • Yes radio is a good approach. Running thru holographic sonar cryptography is a good way to protect the content of your data stream but it WILL NOT protect your internal network, mind you MANY of these systems act as BRIDGES and not routers/switches.

    • Re:Radio? (Score:2, Interesting)

      by sandgroper ( 145126 )
      Could this be used to secure wireless networking?


      Secure??? Who knows. What it should allow with radio is something I've been calling "Space Division Multiple Access". In effect, using scatterers in the environment (e.g. buildings, mountains, what have you) the "cell size" could be brought down to a few tens of meters using the same number of base-station transceivers as currently exist. Who needs more spectrum when you can focus the same bandwidth on multiple physical locations?


      BTW, the New Scientist article is talking about kinda old work. NS had a blurb on this back in '97 or so.

    • Re:Radio? (Score:2, Informative)

      by jlseagull ( 106472 )
      i've done some work on this. this is indeed possible, though the radios on both ends need high sample rates if the communication will be recieved over short distances, which isn't practical on 802.11b cards - the sample rates are in the Gs/s range. in addition, the signal environment that 802.11b operates in is highly variable, and subject to reflective and refractive variations in power on the order of +/-3dB over 10us. Phase variations can be as large as 1000%(that is, 10 times as long as the wave itself) over the same timescale, making phase correlation and interferometrics totally useless. Something as simple as a fan running can perturb the signal environment on the other side of the building to this degree(believe me, we tried it). it might, *might* work over larger static environments like a city or a mountain range, but 802.11b isn't spec'd for that kind of range. so the short answer is, no.

  • Interesting, but too scientific? (Score:3, Insightful)

    by friday2k ( 205692 ) on Wednesday October 24, 2001 @03:52AM (#2471025)
    I think the idea Edelmann is pursuing here has some very interesting implications but also limitations. I wonder how stable the environment on greater distances might be, current, the seabed itself, and other environmental influences. The same goes for the suggested idea of using ping times and number of hop points to encrypt a message. These are highly unstable factors and in order to encrypt the message the environment shall be the same for both sides for the time of the communication flow. But I am also not enough cryptographer to really tell. Maybe others can shed some light on this?
  • The internet lag times on each leg vary from moment to moment, so there's not the same degree of certainty that the speed of sound in water has. This probably wouldn't work. Plus, we've got asymetric crypto, which works very well, thank you.

    Also, in the sonar field, would it be possible to guess at the location of a recipient by catching some of the signals? One wouldn't want to give away the location of your subs, would one?

  • Doubt that it would be useful.. (Score:4, Interesting)

    by Bowie J. Poag ( 16898 ) on Wednesday October 24, 2001 @03:54AM (#2471036) Homepage


    A well-seasoned network admin friend of mine and I once had a conversation over dinner about an idea I had brewing -- An application that would attempt to guesstimate where you were on earth based on triangulating distances from known servers by means of measuring ping time. A small network database that contained, say, a hundred servers nationwide that constantly maintained a list of ping times to a hundred other machines would provide enough coverage and enough data to allow a single machine to guesstimate where it is on earth based upon simple trig.

    The only problem with this idea is that A) Network latency times can change erratically from moment to moment, and B) Some nodes may even drop out of the network due to upgrades or flaming death. Depending upon how fine-grained the mesh is, and depending how accurate you want the guesstimate to be, you could be reasonably certain of at least being able to determine your location within a couple hundred miles.

    Not useful for you and I, I know.. But it would be kinda cool if people could buy PCs, set up them straight out of the box, and the box goes out on the mesh and figures out where it is in the U.S., and sets the time accordingly, suggests local IPs, other stuff.

    Amazing what you can discuss over a bacon cheeseburger, eh?

    Cheers, and yes, PROPAGANDA is still up,
    • With error correction and a means of continual
      adapting to the current situation it would be
      definitely doable. The bandwidth may be poor
      though.

    • > An application that would attempt to guesstimate where you were on earth based on triangulating distances from known servers by means of measuring ping time.

      I can reliably locate Slashdotters in meatspace by observing the time it takes for them to accumulate three troll responses to their posts.
    • Its been done. Read `the cuckoos egg` by clifford stoll. They worked out the hacker was in Germany via a similar method to the one you described.
      • They worked out the hacker was in Germany via a similar method to the one you described.

        No, all they did was measure the time it took his packets to get from place to place, then performed a back-of-the-envelope calculation to guesstimate a distance. It was about as scientific as Dianetics, even if the final answer happened to be more or less correct (as in, within a couple of thousand miles, in some direction.)

        Had the hacker been sitting on the end of a modem in France, but dialed into a machine in Germany, their "system" could have produced an even more bogus result.

    • or even whether it is in the US.

      Finally, a way to get rid around the horrible US_orientated software!

    • Much more reliable a method would be to use a traceroute and look at suffixes until a 'listed' suffix appeared. Just set it up to trace to a few different hosts, and see where the routes begin to diverge.

      This has been quite useful for air based wireless-

      The theory behind it is even a standard part of amateur packet radio. When your using typically 50 watts (or even 1500 watts, legally) you tend to connect to some interestingly distant stations that you'd have no idea where they were if they didn't leave a little identifying information in their 'hostname'

      Ah, yes. Manual routing of packets. Really makes one appreciate all the neat tools we use now..
    • Or you could ask the user for his area code / prefix. Which you probably did before you connected to the net anyway.
    • Check out xtraceroute [chalmers.se]. It gives you a view of the globe (using OpenGL) and will attempt to locate a given IP and the route to that IP and plot it. It uses a list of know routers around the globe and simple rules similar to: If domain name contains .md. then its probably in Maryland...
    • Might work most of the time, but there are very obscure times when it will fail miserably. I used to work at a company in Houston. Our parent company was in California. This was before the time of widely deployed firewalls, and our internet access was through our parent company.

      If I wanted to send email across the street, the email first went to the parent company in Cali, and then across the street. This was true of pings also.

      So, any amount of probes that you had deployed would have thought that my network was about 2500 miles away from California, and none of them would have thought that I was in Houston.

      $.02.
    • There has been a commercial solution out for a very long time, called VisualRoute [visualware.com]. I used it for a job I did a while back. Pretty slick stuff. You really don't need *any* centralized server, if the end box is up - you can find out pretty close where it's at. One of the major problems with this is AOL because all their IPs are divied out of Ohio or some other state (can't recall).

      Sorry ya got beat to the punch, but you can go punch your friend because there is a company that is making a lot of money off that idea.
  • Every time my phone beeps to alert me that "Voice Pricacy is not active" I wonder who could be listening.
    It seems like an approach somewhere between the holographic approach, and the web 'node' approach could be applied to digital/PCS/cell/mobile phones. Does anyone know about research being done into voice privacy on mobile phones?
  • While underwater encryption is a nifty idea, I would much rather we discuss the US government plans to start using powerful sonar communications that, in test runs, have caused whales to beach with under highly atypical signs of death (the equivalent of bleading ears).

  • net encryption (Score:2, Interesting)

    by ruppel ( 82583 )
    Supposing one intercepted the signal underwater it could still be decrypted. Admittedly this would require formidable computing power since one would have to simulate the geometry of sender and reciever in a continuous medium.

    In communications across the net this kind of playing around with different routings and time delays would not be as effective since once intercepted the decoding would be assuming a descreet medium (only so many different pathways). It isn't clear whether the effort put in this kind of scheme would be worth it, ie. it could bne much more effective to refine the encryption algorithm.

    One should note that in descreet systems, like electronic locks that open when a transmitting key is waved in front of it, the principle of asynchronous signaling is already in use. These systems use clockless processors to make the recording and decoding of the transmitted signals near impossible.
    • Supposing one intercepted the signal underwater it could still be decrypted. Admittedly this would require formidable computing power...
      From the article:
      The system works by broadcasting messages in such a way that they can only be received at one point in the water - so no one else can intercept them
      The signal is uninterceptible, not encrypted. The only place in the water where the multiple split signals coincide is the destination.
      • Hmmm.... But it IS interceptable! All you need to do is have an array (or matrix) of listening devices near the transmitter. Then with (massive?) computing power you should be able to search for the sort of correlation that the transmitters form. Right? Probably would help if you knew the distance to the submarine too.

        My question though is why not just steal the buoy?

        --jeff

  • Impossible. (Score:4, Interesting)

    by bornie ( 166046 ) on Wednesday October 24, 2001 @04:12AM (#2471078) Homepage
    "This holographic approach suggests a method of web encryption using multiple hop paths and ping times to create a message which can only be decoded when received at a specific target node!"

    This implies that all routes are static and no routers ever will go down. It also implies that pingtimes are constant between routers/hosts. Both with are false.

    If the IP of all intermediate routers are used in the encryption (which isn't clear) a change of route will make the current 'key' unusable. Further, the ping-time between hosts/routers vary alot as the use of internet vary and will also make this system unusable. A simple DoS-attack will completly destroy any encrypted data in transit which will make it only more insecure.

    --
    Börnie
    • There's more.

      The holographic system works, if I understand correctly, by integrating signal over many known paths, similar to a QED-style Langrangian. The number of possible paths for the sound to go must be large in order to 'encrypt' the message with sufficient complexity. One can integrate over many (i.e. an infinite) number of paths.

      However, if done in net-space, you have only a small, integer number of paths.. perhaps 10 or 20 at most. This would just mean that you are breaking up your signal into 20 discrete packets that the listener can all find. Then the listener just needs to reconstruct the transmission times for all 20 paths to reconstruct the message. This might be difficult, but not impossible, if we make the necessary assumption that net-space transmission times are predictable.
      • And it is still easy to locate the point where all packet will converge in such way that the message will be plain since all packets has the same destination-adress. With omnidirectional sound that is not possible. This does not take in account that ones uplink (if one is an end host) will receive all packets and will be able to decrypt the message even though is is not meant for them.

        If one want omnidirectionality in net-space one has to exclusivly use broadcast-packets which in this case should be routed indeffinitly. This is not only against several RFCs but are also foolish and will break the net. This still makes it possibly for the uplink for end-hosts to decrypt the message, it is not hard for that computer to calculate the result of the last hop for all relevant packets.

        I don't want to see those broadcaststorms if this is used in a large scale. :)

        --
        Börnie

  • heh (Score:1)

    by newt3k ( 128812 )
    kewl, next time i fart in the pool, i'll have to try to encrypt it :)

  • Covert Operations (Score:2, Insightful)

    by DMouse ( 7320 )
    So let me get this straight, they are suggesting that a submarine can communicate securely with something else in the water ... by being really noisy.

    I can see that going down a treat when a sub is trying to keep itself invisible.

  • YRU : Your rights under-water (Score:5, Funny)

    by TheMMaster ( 527904 ) <hp.tmm@cx> on Wednesday October 24, 2001 @04:37AM (#2471116)
    will this article [slashdot.org] on slashdot mean that the FBI will now 'tap' the oceans too??

  • One time pad (Score:1, Insightful)

    by Anonymous Coward
    So they've basically reinvented the one time pad, just using the environment as a key...
  • Multiple routes seems to be pretty hard to come by.
    I'm pretty sure huge majority of systems on the net can only send packets to one gateway and don't have any control in the route those packets take.
    • True, if you're using TCP/IP and nothing higher. Shove a simple gateway on top of TCP/IP (or gateways, really, in multiple locations) and you can get the behavior you're looking for, I think.

      I post this link every time something like this pops up. It's an idea I had last summer, I think, that's along these lines. One of these days, someone will actually read it:

      It's Here. [digitech.org]

  • Ideas anyone? (Score:2, Interesting)

    by sperling ( 524821 )
    I'm not a cryptographer at all, but i'm familiar with the basics and quite interested in the logics behind cryptographic techniques. I wonder, if anyone here have any ideas on a scheme that would let us use the routes (assuming they're static) or the pingtimes (assuming they vary very little) to improve security of a communcation channel? Maybe in a setup with 5-6 different computers all working together in a model designed to do key exchange and validations, to let a new computer into the circle.

    If you think in term of a small distributed network with all point to point secure connections established, how can this be utilized to verify the identity of a new participant?

  • is ìt a crucial part of the article that i missed, or couldnt *anyone* just listen in on the conversation from whereever they like and distinguish two different sets of sounds? i mean, the sounds wouldnt be exactly like the ones the reciver gets, but wouldnt they still be able to
    tell the two waves apart? if they can then this is pretty hopeless

    k
  • We could always use a new encryption technology. Although I would expect that the signal would probably need to go through some other encryption system to make it harder to crack.

    It strikes me that this system is almost an 'obscurity' based encryption which we all know is never a good thing :-)

    The technique reminds me of something I read a while back about a 'directional' loudspeaker that could target an individual person in a crowded area (e.g. an airport). It was sort of like 'laser' but using sound waves from different sources which created an interference sound at a certain point.

  • The exact location of a submarine is of the ultimate concern for its survival during the war time. The holographic approach seems to solve the communication problem.... But, I doubt if that will in fact expose the secret location of a sub.

    Decrypting the msg will be hard, but finding out where the constructive/destructive interference zone s are should be much easier... Hopefully, the system won't become a sub location broadcaster.
  • Although it is a fascinating idea, I seriously doubt you could
    use a similar method for encrypting traffic on the present day
    Internet.

    The biggest show stopper will be the lack of reliable source
    routing. Unless you can reliably specify the route the packet
    takes (or alternatively, predict the route), the whole schema is
    unworkable. IP/4 simply does not support source routing to any
    usable degree. IP/6 does IIRC, but even then, I suspect the ping
    times will not be consistant enough.

    Secondly, a serious change will have to be made to the TCP stacks
    as the time interval between the arrival of packets will be an
    important factor in this system. Again, I don't see how you can
    rely on the transit time given the infrastruture of the Internet.
    Don't forget that this infrastructure is what gives the Internet
    it's power.

    Finally, in the Internet scenario (as opposed to the SONAR
    version) this is as about as secure as private key encryption.
    Unless my machine is multi-homed, there's likely to be at least
    one router that sees every packet my machine sees. This is
    fundamentally different to the SONAR version, where you have to
    be a precise physical location to be be able to "hear" the
    transmission.

    Cute idea, but not feasible.
  • I wonder who's going to be the first brainiac (sorry, excessive VC funded mulch) to try and build some form of network using sound and water as the carrier..

    Imagine it in 5 years.. Worldcom advertising "dark water" - buy your unused water now for $$$$, expect high latency!

    I suppose you've got a lot of bandwidth (wetwidth?)

  • Only in the real world (Score:2, Insightful)

    by joe_fish ( 6037 )
    It won't work on the net well, and there are problems with the idea in the real world too.

    In effect the sea floor and positions of sender and reciever are acting as a secret key. They 'encrypt' the messages and you can only decrypt if you know the secret key in enough detail - i.e. you are the reciever, and the working with the sender. However the snooper in *theory* could decode the signal if he knew enough about the sender/reciever/sea bed, and could do some farily complex maths. How complex the maths is says if it will work in practice. But given that computer can model huricanes, I would guess that modeling the sea bed is plauible.

    In the virtual world though all bets are off. The terrain is very mappable, and fairly simple. So if the problems of varing ping times can be worked out the encryption is very easily broken.

    I wonder if the sea bed version stops working if the tide changes.

  • While the system that governs this type of communication may not be as chaotic as say the weather, it definitely should have sensitive dependency on initial conditions.

    Large amounts of packet loss would occur anytime a fish swims through the line of sight. My question is how sensitive is it to such things. My guess is that a minnow could render a message totally useless. I imagine that is what has kept the Navy from adopting such technology.
    • The frequencies used underwater (10s of kHz) and the speed of sound (~1500 m/s) mean that the wavelengths of most sound waves (lambda = f/c, 15 kHz wave has length of 1.5 meters) compared to average fish sizes are much greater than 1.0... Sound waves of whose length is much greater than the characteristic length of an object in its path will just pass by the object instead of being reflected.... In short, minnows have NO effect...
      • Let me clarify:

        a SINGLE minnow has no effect -- however, schools of minnows can have a conglomerated effect which _is_ significant. Other temporary environmental factors can also interfere -- entrained air in the water column (air bubbles), sea creatures that contain gaseous air in their bodies (i.e. "popping" shrimp).
    • okay, 2 semantic points and an apology.

      1. Certainly, the equations governing fluid dynamics are highly non-linear, and predicting currents may be hampered by chaotic behavior, but schools of fish are pretty far outside the conventional definition of chaos.. they're more like interference.

      2. There is no packet loss, as there is no routing/no packets.

      Sorry, I hate being an jerk on semantics.

  • Better article: Scientific American Nov 1999 (Score:5, Interesting)

    by Alsee ( 515537 ) on Wednesday October 24, 2001 @06:28AM (#2471243) Homepage
    A much more detailed (7 pages) article on time-reversed acoustics appeared in the November 1999 issue of Scientific American.

    I pasted the summary below, but here's a link [sciam.com] to the summary just to make it official.
    Time-Reversed Acoustics
    Mathias Fink
    Record sound waves, then replay them in reverse from a speaker array, and the waves will naturally travel back to the original sound source as if time had been running backward. That process can be used to destroy kidney stones, locate defects in materials and communicate with submarines.

    I thought it was so cool that I wrote a program to simulate the effect. It simulates 1 or more waves emitted by 1 or more sources, and records the waves at 1 or more "microphones". It then treats the "microphones" as "speakers" and plays back the time reversal of the recording. At first the screen is filled with chatoic expanding circles, but after a while the expanding (and fading) circles combine to create a CONTRACTING and STRENGTHENING circle!

    I wrote it for my own curiosity, and the code is "dirty". If there's some real intrest here I could dig it out and clean it up a bit.

    • It's probably worth noting that this works fine as an approximation, but that physically almost any medium will exhibit a certain amount of dispersion. That is, you will have not just

      d^2 u / dx^2 = d^2 u / dt^2

      but also a small diffusion term (size mu)

      d^2 u / dx^2 = d^2 u / dt^2 + mu du/dt

      This cannot be run backwards in the way the wave equation can. Essentially, it loses information, which will be evidenced by instability of your numerical scheme.

      Brian
  • The difference between the water and the Internet is that it's possible to be at different places at the same time, and over a peroid of time on the Internet to intercept trafic.

  • "This holographic approach suggests a method of web encryption using multiple hop paths and ping times to create a message which can only be decoded when received at a specific target node!"

    It suggests no such thing, and the post should be updated to reflect this. The way a sonar wave travels through water is so fundamentally different from the way packets move through the net that the comparison is in fact quite absurd. Indeed, the IP protocol in no way supports the kind of controlled packet delivery the poster is assuming.
  • Since when does the Internet considered a particle wave system? 'Holographic packets' sounds more like an invention of Steve Gibson than a method with sound scientific and technical backing...
  • ... wouldn't I have to get a waterproof computer case to do this?
  • Don't confuse waves and particles.

    This holographic sonar communications system relies on the interference patterns of pressure waves in water (sound). Internet packets do not behave like waves, they behave like particles. There is no interference between them, nor are multiple packets ever combined into one packet.

    Quantum effects allow the merging of particle and wave features, but we don't have that sort of technology in place in the internet at this time.

    (Though such things ARE being researched [berkeley.edu].)

    ~ Chris
  • The point of the discovery is that you can send a message, possibly without revealing your exact location. This is not cryptography. There is probably not a lot of (public) research on this subject - it may be very possible to locate a ship regardless. If it is hard to locate a sender this way, the interesting thing seems to be the distance over which this works.

    Even if distances don't go much beyond 10 kilometers, you can still create a buouy that a sub launches, and uses as a message relay. Or launch a few while enroute and leave a relay network behind.

    Now, if and when this becomes a real world application *nobody* will be sending uncompressed, non-encrypted information over the link. The regular public and symmetric cryptography has a very calculatable 'risk' of decryption in it.

    Btw, Like so many others said: the Internet idea is totally bonkers. That won't work.

    • possibly without revealing your exact location

      I think that's the problem: sending the initial ping will almost guarantee detection, both in water and on the internet; the other problems, the changing (non-static) conditions only come to make this worse, 'cause you have to send additional 'pings' when the environment changes enough to make the transmissino incomprehensible.

  • What makes this a viable option for underwater encryption, is that nobody can sample a big area of ocean entirely to be able to reconstruct the "holographic signal".

    But in the internet, it just only obscures your data. Anyone can read it provided it has backdoors in routers in every path you are using. Yeah, it's harder than monitoring a single router, but still possible, so this approach wouldn't give Real Security[tm]

  • Yeah, but wouldn't the transmit time of sound through water be a lot more constant than that of packets through different internet paths?

  • This holographic approach suggests a method of web encryption using multiple hop paths and ping times to create a message which can only be decoded when received at a specific target node!


    I don't think so. Sound travelling through water conforms to well-understood, consistant physical laws. You can accurately predict how long it will take a sound wave to reach a given destination. However, packet transmission time varies unpredictably based on current load, which changes from millisecond to millisecond. With sonar, if a stationary source pings a stationary target, the ping time will remain constant. With TCP/IP, pinging the same address will give highly variable ping times. Since it appears that this technique is highly dependent on timing, an analogous technique isn't possible on a TCP/IP network.
  • This article sounds like hoax to me (and the fact that its on New Scientist only bolsters that suspicion).

    They say the problem with normal transmissions is that they go in all directions. This means they're also bouncing off of lots of surfaces and echoing back at different times, which is why the sonar ping works.

    However, to play sounds back in reverse as claimed in the article, you'd need to be able to send each piece of the signal directionally, towards the area it came from. If you're broadcasting each piece in all directions, then you're still going to get weird echos off of everything else, and thus end up with weird interference. For the first piece of the transmission this might be OK (since you assume non echo'd transmission will arrive first). But then the echos of prior transmissions will interfere with the actual signal in the parts of the transmission that take longer to arrive. Maybe you could try to subtract these out afterwards? But I suspect its not that simple.

    of course this wouldn't be a problem if they could send each piece of the signal directionally, but then if they could do that they wouldn't need this in the first place...

    am i missing something?

  • It won't work for the Net (Score:1, Informative)

    by Anonymous Coward
    You can't emulate wave interference on the net (soon to be .NET). With sound or light you can use wave interference to either cancel or amplify a wave form depending on the frequency, distance, and position. The sonar technology is nothing new. The same approach has already been applied to Light, and the NSA has already investigated using this technology for secure satilite transmissions, and the DoD is rumored to be using this technology for secure land-line connections.

    However, You can not use wave interference on the net because the information is received as a digital signal. The communication devices have no control over the way the data is encoded on a fiber or copper connection, so its impossible to implement this technology for net traffic. At best if you have control access at both end and you create custom hardware between two points you could use this to encode traffic.
  • This article does not suggest a way to secure computer networks in the least... I'll read it again twice, but I doubt I'll find the missing paragraph that someone must have read...
  • Water more or less cooperates and is predictable in terms of transmitting information (in this case sound); a large heterogeneous network is anything but predictable.
  • A very interesting idea, but it has several flaws for use in covert operations as proposed by the the developers;
    • A submarine is generally moving, thus sending a message recieveable at 'only one point' is problematical
    • If the range is significant, there is a good probability that the sound conditions will change significantly during the combined travel times of the reference pulse(s) and the data pulses.
    • Last but by no means least, for this to work over strategically useful distances, the boat is going to have a transmit a fairly powerful reference pulse, which will be detectable by those who the sender would rather not know they are there at all.
    In the near term, this maybe useful tactically, but not over strategic distances.

    Derek L.
    USN Submarine Service 1981-1991
  • I'm not so impressed by this sonar thing, but some aspects reminded me of meteor burst communications, which is just cool:

    High atmosphere ionization trails from micrometeors (of which there are a surprisingly large number every hour) alter the transmission properties of the atmosphere. Given two stations, you do some geometry, and then wait around for a suitable ionization event. When such an event occurs, transmissions will be symmetrically reflected between the two stations as long as the ionization trail has not dispersed too much.

    This fact is exploited by broadcasting a pseudo-random (like a DS/SS chip) signal from a master station. When transmission sites that know the chip happen to pick up an ion-trail reflection of the master signal, then there exists for a short time a symmetric path back to the master, during which buffered data can be burst transmitted. If the trail lasts long enough, bi-directional communications may be possible. The system as a whole exhibits classic spread spectrum properties, including low probability of intercept, resistance to interference, and channel sharing.

    Meteor burst communication is, however, very low bandwidth, but thats OK - most people using LPI communications aren't exactly streaming MP3s; you can get some pretty good milage with a few dozen bytes of text.

    I don't know anything about the power requirements here... anybody have any usefull info, or corrections to my description? Its been a while since I've looked at this stuff.

  • I dispute the premise (Score:3, Interesting)

    by MarkusQ ( 450076 ) on Wednesday October 24, 2001 @12:11PM (#2472605) Journal
    I dispute the premise of the original article.

    Their logic seems similar to that of "whisper" chambers, but they break one of the assumptions when they start sending a steady stream of phase encoded ones and zeros. Now instead of having to reconstruct a complex wave form, all an eavesdropper has to do is:

    1) Listen for pink-noise with a strong 1kHz component.

    2) Play with the (recorded) signal a bit (e.g. adding 1us delayed copies to the original) until you can decompose it into two types of 1us segments--call them A & B.

    3) Now you have a stream of As and Bs, and two possibilities; either A=0 and B=1, or visa versa. Test both.

    -- MarkusQ

  • Oh yes you CAN intercept it along the way (Score:4, Insightful)

    by PD ( 9577 ) <slashdotlinux@pdrap.org> on Wednesday October 24, 2001 @12:11PM (#2472606) Homepage Journal
    Theoretically, at least.

    In astronomy, the coolest research is in adaptive optics (do a Google search and you will be reading in fascination all day). Here it is in a nutshell, step by step:

    1) The earth's atmosphere is turbulent. That turbulence causes the images of stars to dance around in telescopes, making the image all fuzzy. This is what causes the stars to twinkle when you look at them. Avoiding this problem is the big reason why the Hubble Space Telescope gets such amazing photos when it is much smaller than the largest telescopes on the Earth.
    2) How to fix this problem without launching telescopes into space? Adaptive optics, of course. If you can flex a telescope mirror into exactly the right shape, you can compensate exactly for the distortion that turbulence introduces into the image, removing the majority of the noise from the signal. Suddenly the image becomes almost perfectly clear and steady, not fuzzy.
    3) We know that stars look like points of light, even through the largest telescopes. When we receive a fuzzy image, a very fast computer figures out what shape a mirror would have to be to focus that fuzzy image back into a single point of light. That star is called a reference star. Any interesting objects close to that star are also therefore made clear.
    4) Commands are sent to mechanical actuators on the back of a mirror that deform it to the correct shape to focus the reference star. This happens very quickly, so the resulting image is steady and sharp, despite all the turbulence in the atmosphere. Neat trick.

    OK, so that's how it works.
    You can do the same thing to submarines too, if you know what they sound like. The submarine's sound becomes the "reference star" in this case. When you receive the garbled signal, you might be able to correct it based on the sub's sound. If you apply that correction to the message as well, you might be able to hear the message.

    This has a lot of problems, so practically it wouldn't work. For example, the easiest way to defeat the intercept is to change the noise that your sub makes, maybe with a random noisemaker. But that makes your sub less quiet. Also, the person trying to make the intercept would have to be listening to the sub before the message is sent, because once the message is sending, that would make the sub a random noise and you couldn't focus the sound. And, since the turbulence conditions change (I don't know how fast), over time your ability to focus the sound into a message would steadily degrade. The sending submarine would only have to figure out how fast the sea conditions are changing, and only start sending the good parts of the message after you've lost your ability to focus the sound.

    • You can do the same thing to submarines too, if you know what they sound like. The submarine's sound becomes the "reference star" in this case. When you receive the garbled signal, you might be able to correct it based on the sub's sound. If you apply that correction to the message as well, you might be able to hear the message.

      I see a significant problem in using the sub's sonar signature as a baseline, it's sort of obvious actually. The boats are damn quiet. Or at least can and should be. Missle boats, the russian akula, and the Seawolf class are so quiet that the best way to look for them in the open ocean is to look for a "hole of no sound" where you think ambient ocean sound should be.

  • The sonar xscreensaver rocks. It plots hosts on the sonar screen based on ping time.
    man sonar:
    The sonar program displays a sonar scope on the computer's screen. This scope
    polls a sensor as the sweep goes around the scope and displays what it finds as
    bogies on the screen. The program is designed to support different modes repre-
    senting different types of sensors. Currently the only implemented sensors are a
    simulator, and a network ping function that pings hosts and plots the results on
    the scope.
  • Am I missing something, or would this method be completely useless if the submarine is moving in the slightest bit? How do you get a submarine to be motionless? There's going to be currents and such moving it. Won't nearby ships moving throw it off as well?

  • This wouldn't quite work on the web. (Score:3, Insightful)

    by BlueTurnip ( 314915 ) on Wednesday October 24, 2001 @11:57PM (#2476052)
    If one reads the article carefully, one would discover that this "encryption" technique makes use of the wave nature of sound to both obscure the data in transit, and reconstruct it at the final destination.

    There is no analogy for web traffic which travels over IP which is sent as discrete packets of bytes. They resulting packets cannot be made to interfere with each other at the destination to produce plaintext, nor do they interfere and reflect and become distorted in transit!

    The closest analogy would be to split a message into many small parts and send them along different paths in the hopes that no one could catch them all in transit, but then timing isn't really an issue at all as others have suggested. Also, anyone bugging your connection to the internet (your ISP for instance) could still catch all the packets, ditto for the source. Some have suggested splitting keys and sending some parts by snail-mail, others by FedEx, others by e-mail to different accounts which you read on different machines, and that is really a form of security through obscurity, not encryption, whereas the sonar technique is more like encryption in that even if an adversary knew that information was being send and knew from where, they could't recover the plaintext unless they were at the target location.

    Perhaps quantum cryptography is a better analogy to what's going on, but it's not a perfect one either as there are fundamental differences between accoustical waves and quantum wavepackets.

Slashdot Top Deals

A morsel of genuine history is a thing so rare as to be always valuable. -- Thomas Jefferson

Close