Quantum crypto hack smacked
Implementation flaw only: 'perfect security' still coming
A hack against systems running quantum key cryptography only worked because of implementation errors, according to new research.
A paper published in September demonstrated how the avalanche photo-detectors as used in quantum cryptography rigs might be blinded, essentially causing equipment to malfunction without generating an error indicating that a key exchange needed to be abandoned.
Using the ruse, a potential eavesdropper might be able to gain information about a secret encryption key being exchanged over a supposedly super-secure optical fibre link. Normally any attempt to observe the key exchange would interfere with the process, producing so many errors that a key exchange is abandoned.
However, in real world systems, errors do occur and systems might be tuned to compensate for this. The attack subverts these error detection features in such as a way that a potential eavesdropper could extract at least snippets of data from a key exchange while leaving users none the wiser that anything is amiss.
Commercial systems from MagiQ Technology's QPN 5505 and ID Quantique Clavis2 systems were shown to be potentially at risk from the attack by a team of computer scientists from Norway and Germany.
Computer scientists from Norwegian University of Science and Technology (NTNU), the University of Erlangen-Nürnberg and the Max Planck Institute for the Science of Light in Erlangen showed the crypto-busting technique was possible using off-the-shelf (though admittedly expensive, $50K) hardware. The boffins worked with manufacturers to resolve the problem, which revolved around real-world implementation mistakes, rather than anything wrong with the basic principles of quantum key cryptography, which remain rock solid.
Scientists at Toshiba’s Cambridge Research Laboratory evaluated the work, concluding that the detector blinding attack would fail to work against single photon detectors - providing the system was operating quickly. Straightforward adaptations on potentially vulnerable avalanche photo-detector systems would also thwart the attack, as the Toshiba team explains.
The attack is only successful if a redundant resistor is included in series with the single photon detector, or if the discrimination levels are set inappropriately. Furthermore, by monitoring the photocurrent generated by the detector, it is possible to prevent all bright light attacks on avalanche photodiodes.
The Cambridge-based boffins praised their fellow egg-heads, saying the blinding attack only goes to show that quantum cryptography systems are getting a thorough going over, a process that will likely result in better and more secure systems. Applications for quantum key cryptography include banking and government.
Dr Andrew Shields, assistant managing director at Toshiba Research Europe, commented: "Quantum cryptography is now entering a new phase in which the security of particular implementations is carefully analysed and tested. This is important to uncover any security loopholes and to devise appropriate countermeasures."
Shields added: "It will allow real-world devices to approach the perfect security that can be proven for the protocol."
The Toshiba team published their findings in the December edition of Nature Photonics, the same academic journal that published the original Norwegian/German research. ®
Super secure secret message transmission...
... then the data is stored on a computer with a CD burner, with full unmonitored/unaudited access granted to every low-level Lady Gaga-loving flunky in the building.
secure, insecure or a quantum superposition of both
No doubt 'system being tested' indicates the last of these probabilities.
Quantum denial of service
Again, it seems to me that all a malicious person needs to do is "continually observe" a quantum data stream thereby alerting the subscribing parties of an intercept attempt.
Since they couldn't be sure the keys weren't compromised, they would never be able to established secure communications.
I'm curious to know how this apparent weakness will be foiled in operational systems.