Australian Gov't May Employ a Homegrown Quantum Key System

mask.of.sanity writes "The Australian government is trialling a new Quantum Key Distribution (QKD) system built by Aussie scientists. QKD is considered the world's toughest security because the slightest attempt to intercept the one time keys, coded into lasers at the quantum level, will disrupt the beam. The technology differs from current cryptography tech primarily because it's cheap. Well, less than the $US100k price tag of rival systems. It uses off-the-shelf networking gear instead of proprietary technology, and is built on open standards, so it's easier to install. The random key is encoded at the quantum level in the sidebeam in the phase and amplitude, or brightness and colour, of a highly tuned laser beam. The creators, who built the system in part for their Ph.Ds, said it can be used to transport the most sensitive data like critical infrastructure and secret commercial IP. The days of hand-delivered security keys are numbered."

Re:Wait a minute...

[wdsci]

Last I heard, quantum cryptography did require a dedicated line. And you can't intercept and regenerate the signal because the laws of quantum physics make it impossible to measure enough information about the beam to generate a copy of it. The way quantum cryptography works (at least this is one simple scheme), the sender of the key transmits photons that are polarized in one of 4 directions: N-S, E-W, NE-SW, or NW-SE. But when you measure the photons, you have to choose whether to make a N-S vs. E-W measurement, or an NE-SW vs. NW-SE measurement - you can't make both. And if you choose the wrong one for any particular photon, the outcome of the measurement is random (and the original orientation of the photon is lost). Although, the sender and receiver of the key will have to compare notes via non-quantum means, to see which photons they measured using the same scheme, and if you have access to both the quantum channel and the non-quantum channel, I guess you could pull off a man-in-the-middle attack.

Re:Wait a minute...

[shadow_slicer]

The key is not encoded -- it is random. Both the "sender" and receiver have no idea what the photon's characteristics are. They both flip coins to see which type of measurement to make. Then they keep the bits where they made the same type of measurement and throw away the others.

Any intermediate party will either receive the photon (so the receiver won't) or not receive the photon (and can't measure it). Further, no intermediate party knows what measurements the sender and receiver will make so they can't make the same measurements. If the intermediary can't make the same measurements then it can't generate the same key, and can't generate a passable photon for the receiver. Assuming the sender and receiver have another channel which is secure against man in the middle attacks (though not necessarily secure against eavesdroppers), they can tell each other which type of measurements they made and know what to keep.

Re:Wait a minute...

[bh_doc]

I suspect we differ on the definition of "intercept". If you strictly mean "capture and extract information from", then I agree. Any measurement (the "extract information" part) will collapse the wavefunction, destroying the quantum coherences and ultimately (with approaching-unity probability) being detected by the QKD scheme. However, I was using the term in the more general sense of "have some device between", in which case what I said is entirely correct. Here's why:

Sure it's possible to intercept the quantum signal, but it is not possible to regenerate it precisely - by which I mean reproducing the original quantum state. Read up on the "no-clone theorem" - for example Wikipedia's article.

I'm well versed in the no-cloning theorem. As such, I know why it doesn't apply here. The no-cloning theorem is in relation to making an identical and independent copy of any (a general) quantum system whilst retaining the original system. In this context it would amount to producing a duplicate signal, independent but equal to the original signal. This is not possible under the no-cloning theorem. (I'll preempt a point here, too: Entanglement is not cloning, although it can sometimes look similar.)

But, intercepting and regenerating the signal does not necessarily involve ever having both the original and regenerated signals existing at the same time. Take an example of a kind of quantum repeater, a device that converts a photon signal into some other quantum state, say electron spin, and then converts that spin into a new photon signal. It's roughly the same idea as classical repeaters in long-distance fibre-optic communications. Now, I consider this operation to be an interception of the signal and generation of a new signal with the same information. It's a coherent process; all the quantum information in the original signal remains intact. But you can't get back the photons from the original signal, so the no-cloning theorem is not relevant. (A more detailed explanation of the workings of a quantum repeater could include entanglement, which also means no-cloning theorem is not relevent.)

A restriction on the device is that, to function, it cannot collapse the wavefunction. That means that (at a minimum) it cannot make a projective measurement of the quantum state. Thus, it cannot make any recorded measurement on the state, because that would require making a projective measurement, which would require defining a projection basis (randomly(!), because there's no better way), which would collapse the wavefunction, which would rightly end up being detected by the QKD scheme as eavesdropping.

So, you can have a device which intercepts and regenerates the signal, you just can't ask it any questions.

Re:Okay but why?

[Cimexus]

Well in Canberra (capital city of Australia), most government departments in the Parliamentary Triangle (where all the major/important Government departments area) are connected by a such a dedicated fibre network, that is completely physically separated from the Internet and other public networks This is particularly the case in the defence/intelligence precinct (which is a cluster of buildings in one particular suburb).

Interestingly I tried Googling it and couldn't find much at all. But it exists ... I've used it myself as a contractor to several AU Federal Govt. departments. So you could use this kind of encryption on a network like that I imagine.

But yeah, this technology seems like it wouldn't have huge application outside of these rare, special types of networks.

Re:one time pad

[OzRoy]

This *is* a one time pad. This is Quantum Key Distribution. The quantum part ensures your key has not been intercepted. Once the key is recieved by the other party the actual message is encrypted by using one time pad.

So the advantages of this over a courier is it's a hell of a lot faster and you guarantee the courier hasn't been mugged and had the pad copied.