Practical Quantum Cryptography

Alien54 writes: "Saw this on Eureka Alert: A commercially available system for quantum key distribution and cryptography has been released by a spin-off company from the University of Geneva. The system has been tested over distances up to 70km (from Geneva to Lausanne) through standard optical fiber cables and connects to PCs via USB ports. Transmission rates of about 60 bits per seconds were achieved, sufficient for key distribution. You can see the original abstract here, which also has a link for download of the paper in PDF, Postscript, and other formats."

Comments

Re:All I have to say "neato"

[einer]

I'm obviously missing a great deal of information. I thought quantum cryptography relied on quantum computing, which as of now, doesn't exist. Anyone care to fill in this (egregious) gap of knowledge?

QC solves confidentiality, but not authentication

[The Pim]

I've heard it said that, if QC proves practical, the code-makers will have a final victory over the code-breakers. This seems true at first: there is absolutely no way, in theory, for anyone but the recipient to receive the message.

But how do you know who the recipient is? QC offers no authentication. If you have to use public key for authentication, what advantage is gained by using superior methods for confidentiality?

The only one I can think of is that, with conventional cryptography, you can capture the data stream and crack it "off-line". I suppose that this is significant: with QC you only have to worry about whether they've cracked your private key (that you will use for authentication) already, not whether they can in 100 years (because you've used it for encryption).

Another argument might be that it is easier to eavesdrop on a channel than to redirect it. But that seems like a dubious assumption, if the enemy is determined.

Thoughts?

Re:60 bits/second doesn't seem useful

[mike_g]

The 60 bits/second is only used for the key exchange. The key is for a ordinary symmetric cipher, such as RC5, DES, AES, or Blowfish. For example a 64 bit key for RC5 would take slightly over a second to transfer. And our friends at distributed.net have been working on breaking a 64 bit RC5 messages for the past four years and still have not exahausted the keyspace. Symmetric ciphers have many more valid keys per bitsize than asymmetic ciphers, giving you stonger protection. Keys for symmetric ciphers don't need to be as large as those for asymmetric cipers, 60 bits/second is a little slow, but definitely adequate.