There are many diagrams out there explaining how quantum key distribution works on a physics /optics and mathematics level. I understand these explanations fairly well so I'm not asking about the underlying principles. This is one of the few that actually features people on it (although they seem to be fairly incidental to the usage):-


As I understand it, the machines only transmit keys not actual business data. I appreciate the exchange is done with one time pads, but is it like Diffie–Hellman key exchange otherwise, but only between fixed machines /servers? Do they replace entering AES keys transmitted over the phone /fax? Since AES is hard to break, surely the keys don't need replacing every hour? And do they then operate in pairs, because two is only a network from a marketing perspective. They can't control the log in into my desktop computer can they? And wouldn't all the desktops have to be interconnected with fibre optics rather than Ethernet? The latter is probably a rather naive question.

This additional image clearly shows the transmission of secure encryption keys through a Toshiba quantum network:-


At a separation of 20Km, the devices are generating /exchanging keys at a rate of 10 kbits/s. Why as that's 40 no. 256 bit AES keys each and every second? What exactly does someone do with 140,000 unbreakable AES keys per hour? Even more ridiculously, that's a million new keys in a single business day.

I'd appreciate an answer from a operational perspective, although I realise that this is firstly a cryptography theory site. So what's it for?

I looked at Quantum key exchange skepticism/confusion and a simulated network, but both were unfruitful.

  • 1
    $\begingroup$ That machines do not transmit keys. They produce. Yes, keys could be used for AES or as a one-time pad. $\endgroup$ – Vadym Fedyukovych Jun 7 '17 at 7:51
  • 4
    $\begingroup$ My layman's understanding is the quantum channel is like a pipe that allows the key to be sent from Alice to Bob, but if someone taps into the pipe to listen in it changes the quantum properties of this channel and allows them to detect the tap. So the difference between using this special quantum channel is that if you control both ends you can be sure there are no eavesdroppers. $\endgroup$ – daniel Jun 7 '17 at 19:26
  • 1
    $\begingroup$ @Paul Uszak It would be counter-productive to talk about another "multimedia" definition. It would be better to learn BB84 protocol. $\endgroup$ – Vadym Fedyukovych Jun 7 '17 at 20:43
  • 3
    $\begingroup$ @VadymFedyukovych Why do you feel the fact that you are not transmiting keys but rather creating them is important? In the end it's a key distribution channel and given the fact that currently the channel is pretty much point to point using a special channel it's mostly a gimmick which adds little over distributing a couple flash disks with random data. $\endgroup$ – DRF Jun 8 '17 at 13:18
  • 1
    $\begingroup$ @DRF Please forget about transmission of keys of any kind. This is not true. Measuring a quantum state is the important difference. $\endgroup$ – Vadym Fedyukovych Jun 12 '17 at 19:03

It seems that the initial diagram is correct. A QKDN is exactly this:-


and it relies on two pieces of kit, a quantum key generator and an encryptor device. A 256 bit AES key is generated and shared between Alice and Bob via the BB84 protocol running over dedicated fibre optic cable. The key itself is created via a true random number generator (perhaps based on the phase interference of a laser beam).

The keys are then passed to the encryptor devices which operate AES between themselves over conventional channels. So actual business data flows only between the encryptor devices, over common networking technology and not the quantum machines. The thing that ensures the security of the transmitted data is that the AES keys only have a life span of one minute. They're then replaced. That's why all that quantum data is transmitted. AES keys are constantly being replaced before anyone could even attempt to break them. Clearly some of the numbers here are dependant on individual networks and hardware.

And QKDNs are used in anger all over the world including government and commerce. The first was deployed in Switzerland to secure their frequent public votes /elections. And there are some interesting future developments that won't require fibre optics at all, and some that will ensure secure key exchange over infinite distances.

All of this is to be found on YouTube. Don't laugh. It's a presentation by Kelly Richdale, head of quantum randomness at ID Quantique, a market leader in QKDNs. There is also an unbiased and fair summary of QKD from the European Network and Information Security Agency (ENISA) here. It's concise but rather dated (Nov 2009), so a lot of the open issues should have been resolved or are in the process of being so.

| improve this answer | |
  • $\begingroup$ Yes, BB84 exactly. Yes, "key is created" is better than "transmitted". "Based on quantum measurement" would be even better. Lots of relevant lectures could be found on Youtube. This presentation starts with "background is not quantum physics" and with "MBA". To learn BB84, Norbert Lütkenhaus' lecture youtube.com/watch?v=dh3ij4ANsZ8 could be better. $\endgroup$ – Vadym Fedyukovych Jul 6 '17 at 22:29
  • $\begingroup$ I really hope it doesn't take off, otherwise cryptography is going to become very boring. $\endgroup$ – Melab Oct 20 '17 at 1:59
  • $\begingroup$ @Melab Sir, I think that you've revealed the true reason that QKD is dissed on this site. It's nothing to do with security. It is likely to make cryptographers irrelevant and hand encryption over to physicists and engineers. Even children can use one time pads. It think that it's a black cab versus Uber situation and it's just Luddite behaviour. $\endgroup$ – Paul Uszak Oct 20 '17 at 10:26
  • 2
    $\begingroup$ Quantum key distribution doesn't make cryptographers "irrelevant" anymore than science renders philosophy "obsolete". $\endgroup$ – Melab Oct 20 '17 at 16:44

I would say the quantum key distribution networks being built right now are for research. Comparing this research to other fields they would be at the 1969 level for space exploration and for the ARPA net. But we don't know yet if it will turn into something everyone will use for everything, like the internet or something that only has limited potential like space exploration (which only gave us GPS and spy satellites).

The main sales point is instead of making a encrypted communication channel a mathematically hard problem to solve like RSA, you make it an impossible physics problem that can't be solved.

| improve this answer | |
  • 1
    $\begingroup$ 'you make it an impossible physics problem that can't be solved'; two problems: a) that ignores side channel attacks on QKD systems (a number of which have been demonstrated), and b) if you use the QKD to generate symmetric keys, like the illustration in the question, then the system is no stronger than the symmetric system; which raises the question: if you depend on AES to encrypt your traffic, why do we depend on QKD (rather than an AES-based key distribution system); what advantage does it bring? After all, QKD has significant costs; does it bring significant benefits? $\endgroup$ – poncho Jun 8 '17 at 15:54
  • $\begingroup$ @poncho on b) its replacing RSA in the SSL type of setup, which is the first to go once quantum computing is developed, also I would use OTPs not AES, because I can't understand the math for AES. On a) side channel attacks, the concept is nice even if some previous implementations have been side channeled, do these attacks need physical access at one of the ends or do they work in the middle? $\endgroup$ – daniel Jun 8 '17 at 16:06
  • 1
    $\begingroup$ As for SSL, I don't see how QKD can work. I do not have a dedicated fiber link to an amazon server, nor is one in line-of-site. How could I then use QKD to SSL-connect to amazon? If we assume an expensive on-premise box (which QKD assumes), why would be inferior if that box used AES (or OTP) to do key exchange instead? It would certainly be cheaper, and more flexible (don't need a dedicated fiber link). As for side channel attacks without endpoint physical access, yes, they have been demonstrated. $\endgroup$ – poncho Jun 8 '17 at 16:22
  • $\begingroup$ @poncho I'm not saying we can right now jump on amazon via QKD, I'm saying they are researching QKD now, possibly to replace RSA as a means of distributing keys, since RSA might have problems holding up against more developed quantum computers in the future. $\endgroup$ – daniel Jun 8 '17 at 16:31
  • 1
    $\begingroup$ @floorcat One of you practical & robust examples might make an excellent answer to end all this - err ... uncertainty. $\endgroup$ – Paul Uszak Jun 8 '17 at 22:25

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.