Quantum communication systems offer the promise of virtually unbreakable encryption.
Unlike classical encryption, which secures data over networks today and whose security depends on the difficulty of solving mathematical problems like the factoring of large numbers, most quantum encryption schemes keep the encryption key separate from the data.
This approach ensures an eavesdropper with access only to the data could not decipher the key. However, researchers now know even quantum encryption may be susceptible to hacking.
There are now new ways to calculate the failure probability of certain quantum encryption schemes, and Renato Renner of the Institute for Theoretical Physics in Zurich will discuss how he and his team of theoretical physicists will show who in a presentation next month at the Conference on Lasers and Electro-Optics (CLEO: 2013) in San Jose, CA.
The numbers would allow users to estimate how likely it would be that an attacker could read their secret messages — information critical for ensuring the overall security of quantum communications.
Quantum key distribution (QKD) is a quantum encryption where a secret password end up shared between two distant parties (usually named Alice and Bob in thought experiments). The secret password, or key, distributes as bits of quantum data, so if an eavesdropper (usually named Eve) tries to intercept the message, the bits will end up disturbed and Alice and Bob will know the transmission suffered a compromise. If the key does not end up disturbed, it can encode messages sent over an insecure channel.
“The security of Quantum Key Distribution systems is never absolute,” Renner said. He notes that the security of QKD systems depends on three assumptions: The initial secrecy of the password, the correctness and completeness of quantum theory, and the reliability of the devices in the quantum communication system.
Recent work by other research groups illustrated how real-world devices not 100 percent reliable can leave weaknesses in quantum communication schemes that may suffer exploitation by a clever hacker. For example, the photon detectors used in QKD should click with a certain probability whenever it detects a photon, but in practice the devices can be “blinded” by a strong light pulse and not click. “In fact, an adversary may use strong light pulses to ‘remotely control’ the detector,” Renner said.
Since such bright light hacking techniques first came out in 2010, physicists have been keen to find ways to calculate the security of quantum encryption schemes without making assumptions about the reliability of the devices. The quest has generated a lot of interest in a field called device-independent cryptography.
“In device-independent cryptography, the proof of security is based solely on directly observable correlations between sender and receiver, and it does not matter how these correlations have been established,” Renner said. “Even if the detectors were blinded, for instance, as long as they produce the right correlations, a secret key can be extracted from them.”
That differs from the traditional approach to calculating quantum encryption security, which is only valid in the nearly impossible case of the devices working exactly according to theoretical specifications.
Renner and others are working on theory-based calculations that establish the device-independent security of certain QKD systems. “With modern proof techniques, it is now possible to quantify their security in terms of a ‘failure probability,'” Renner said. “Specifically, it is possible to make claims such as ‘the probability that this particular QKD system can be broken is at most 10-20,'” which is a very small number.
Renner said it is important to be able to reliably calculate the order of magnitude of the failure probability of an encryption system, whether it is tiny like 10-20 or significantly larger. “Compare it to an aircraft,” he said. “Once we realize it is not 100 percent safe, we want to be sure that the failure probability is still small enough so that we are ready to carry the risk. If we have a system that may fail, but do not know how likely it is to fail, then we will probably not want to use it.”