As far as I know QKD needs a quantum channel to send the photons. If anyone measures the photons they are useless.

It sounds as if Alice needs a fiber optic cable all the way to Bob - or at least line-of-sight to Bob. So no wifi through walls.

One of the things that really makes the internet work is packet switching done in routers. If every sender would need a cable to every receiver then the internet would never have been such a success.

How would you build routers that can route photons from Alice to Bob for QKD? Will they work, if Bob is using his tablet under a blanket on a cruise in the middle of the ocean?


I have attended a few talks on quantum computing breaking currently used asymmetric ciphers. Usually the presenter then mentions that quantum encryption will solve this. The presenter mentions this as if we can use the existing infrastructure, which I feel certain is wrong. So I am wondering if I am missing some obvious way to be able roll out QKD to every computer.

  • $\begingroup$ "Usually the presenter mentions that quantum encryption will solve this." - Are you sure it's not "Usually the presenter mentions that post-quantum cryptography will solve this."? The two are very much not the same, and the latter is more widely regarded as a serious solution to a problem in the relatively near term, that works on existing infrastructure (which you can clearly see is a practical problem for quantum cryptography) $\endgroup$
    – Ella Rose
    Dec 9, 2017 at 17:26
  • 1
    $\begingroup$ @EllaRose: "are you sure it's not 'Usually the presenter mentions post-quantum cryptography'"; that depends greatly on which talks he is listening to. There are a number of talks given around QKD, and they rather ignore any other solution... $\endgroup$
    – poncho
    Dec 9, 2017 at 18:12
  • $\begingroup$ @EllaRose Yes I am sure. Some of them mention pqcrypto in passing as a temporary solution before we get to quantum encryption. Personally I think pqcrypto has much more practical use than quantum encryption, and my question is trying to uncover if I missed something. $\endgroup$
    – Ole Tange
    Dec 9, 2017 at 18:39

1 Answer 1


How would you build routers that can route photons from Alice to Bob for QKD?

I know of three possible ways of addressing this problem (only one of which is currently practical).

  • Have the routers have QKD systems on each link; and so Alice does QKD with router 1, which does QKD with router 2, which does QKD with Bob. The obvious issue with this is that router 1 and router 2 also get the keys. This obviously doesn't work on the Internet (which you don't trust every service provider between you and Bob), it may be plausible in, say, a data center (where the same organization runs the entire network, and so does have some trust on their own infrastructure.

  • Have routers that don't actually measure the quantumsignal. I have heard of two proposed subvariants of this one:

    • One based on mirrors that can be moved ultraquickly. In this case, when Alice wants to talk with Bob, she informs the routers, which would quickly set their internal mirrors so that photons from Alice get reflected at Router 1 to Router 2, which reflects them to Bob. Now, this doesn't do anything about the QKD length limitations, and I actually haven't heard this option mentioned in the last several years, and so people may have given up on it.

    • One based on Quantum Amplification. In this case, Router 1 would take photons from Alice 1, and in a way that doesn't break the Quantum State, forward them to Bob in some amplified/quantum error corrected state. We currently can't do this; this appears to be quite difficult (perhaps not much easier than building a full Quantum Computer), and I'm not exactly sure how they intend to get around the 'no cloning theorem' (which states that you cannot duplicate a quantum state, hence it's unclear how they expect to amplify one).

  • QKD satellite; here, there is a satelite that sends pairs of entangled photons to the two end points (and the end points use their entangled photons to agree on keying material). There's just been given a proof-of-concept, and there are several obvious issues ("does everyone have a satelite receiver? How do you talk between end-points that are beyond satelite range? How do you deal with satelites that can communicate only intermittently (currently: at night, when the weather is clear, and when the LEO satellite is over the horizon)?"

  • $\begingroup$ If I control the QKD satellite, can I then send photons that look like entangled photons, but for which I can predict what they will measure? In other words: Is the satellite solution safe, if the satellite is run by the devil? $\endgroup$
    – Ole Tange
    Dec 9, 2017 at 19:18
  • 1
    $\begingroup$ @OleTange: good question. The answer is, counterintuitively enough, yes. Things such as Bell's inequality can be used to distinguish between 'truly entangled states' vs 'states that happen to be the same'. This is a situation where classical mechanics cannot describe the situation... $\endgroup$
    – poncho
    Dec 9, 2017 at 19:54

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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