How it works
(Courtesy of the Australian National University)
Vikram Sharma explains how the group's quantum key distribution system works. In classical cryptography, the sender is referred to as Alice, the recipient as Bob, and the eavesdropper as Eve.
A laser beam has billions of photons travelling per second. We call this the carrier frequency. We make sure that beam is as precise and noise-free as possible. From the carrier frequency, we shift a small number of photons out of the billions, and drop them at frequencies that are displaced, plus or minus, from the carrier frequency. In this way, we encode the random number ones and zeroes on the side bands of the laser. Using this method we've setup an experimental prototype to simulate transmissions over 50 kilometres of fibre.
Bob's receiver station makes very sensitive measurements of the laser beam to strip out the carrier and look at what was coded on the sidebands. This gives him the raw key, but the encoded data has been jiggled around by quantum noise. The key that Bob has received differs from that which Alice has sent. Also, in quantum cryptography we are totally paranoid, so we always assume that Eve is omnipresent and omniscient. We assume that all the losses in transmission are somehow captured by Eve and she is able to make optimal use of this information.
The raw data that Bob has received is manipulated in various ways to adjust for characteristics of the electronics and the transmission channel. We then carry out a procedure called post selection. This involves capitalising on the random nature of quantum noise by making the most of those events where Bob received more information than Eve. The idea is for Bob and Alice to capture those transmissions where Bob did really well in his measurements, and discard the rest.
Via a process called advantage distillation, we amplify the advantage that Alice and Bob have eked through the post selection procedure. This comes at the expense of discarding more bits of the raw key.
We then engage in a process of secret key reconciliation, which is essentially a procedure for error correction. Recall that the data is noisy. Bob makes his key match Alice's through a series of indirect disclosures that are somewhat analogous to the game Master Mind. But if Bob discloses information about his bits, even if it is indirect, then an eavesdropper could also get to know this information. Again we assume that Eve is eavesdropping and is trying to maximise her knowledge of the secret key using a composite of all the information she has gained from all her eavesdropping activities.
After the key reconciliation step we've got a perfectly clean key. The final stage is to negate the usefulness of all the information that Eve has. We draw on abstract algebra to carry out an operation that is somewhat analogous to kneading dough. Imagine that the information that Alice and Bob share is represented by a piece of pizza dough. We assume that Eve has gained considerable knowledge of the key as well from her eavesdropping, however she still has some uncertainty about some of the bits. If we represent those bits that she doesn't know anything about by a dot of blue ink on the dough, by kneading it we spread the ink throughout her dough. Alice and Bob don't have a problem because their bits have been perfectly aligned prior to this operation - even if you jumble them, it doesn't matter. And just to be sure that she can't reverse engineer the process, we chop off a bit of the dough and use only this small portion for the final key. Even if Eve had some incredibly powerful computer, she'd never be able to reconstruct Alice and Bob's final secret key.
Vikram Sharma can be contacted at email@example.com