Alice and Bob are foiled once again.
Image: Nayu Kim/Flickr
The problem with the theoretically perfect security of quantum encryption is that, like anything else in computing, it depends on hardware: physical machines.
It's much, much more difficult to make a perfectly secure or incorruptible physical machine than it is to make a perfectly secure algorithm or theoretical system. An example of this comes in the form of a Trojan horse attack in which a focused beam of light is fired into an encryption device and the reflection is used to piece together how the previously sent quantum key was created and, thus, what the key actually is or how to recreate it.
Last month, a group of researchers offered a passive security system (the first of its kind) designed to beat just such an attack. Almost as quickly, however, encryption researchers based at the University of Waterloo's Quantum Hacking lab have come up with a novel way of ensuring the success of a quantum Trojan using high-power laser beams capable of burning an observation port of sorts right into a key-generation device.
Their work is described in a paper posted recently to the arXiv open-access pre-print server.
First, a quick quantum encryption refresher. We start with Alice and Bob, the sender and receiver of a secure message. The security of this message depends on one of the defining features of quantum mechanics, which is that it is not possible to observe or measure a quantum system without disturbing it in some way. Thus, if the message prepared by Alice, which is represented by some quantum state, gets hacked somewhere in the middle, Bob will immediately know based on the disturbance and some extra side-channel info from Alice.
The Trojan attack works because it doesn't interfere with the message and instead targets the message preparation set-up, which should yield the same information.
"In classical [non-quantum] communication systems, the security-critical parts can be physically separated from the communication channel, thus making them isolated from physical access and alteration by the eavesdropper," the Waterloo team explains. "However, the front-end of a quantum communication system is essentially an analog optical system connected to the channel, and easily accessible by an eavesdropper."
"The damage opens up a new side-channel, which can compromise the security of QKD even with today's technology."
The group's counter-encryption scheme beats anti-Trojan security measures by not just probing the encryption source, but changing it. Rather than fire a laser back down the message path at the encryption device, their system uses a high-power laser to burn a tiny hole in it, which can then be exploited via a more traditional Trojan attack.
Thus, the system can be rendered insecure.
"To verify this possibility, we perform laser damage on two completely different widely used implementations: a commercial fiber-optic system for QKD and coin-tossing with phase-encoded qubits, and a freespace system for QKD with polarization-encoded qubits," the Waterloo group writes. "In both systems, the damage opens up a new side-channel, which can compromise the security of QKD even with today's technology."
There are still some countermeasures available to quantum encryption protocols, but none are particularly simple or passive. The main possibility is "hardware self-characterization," which sort of just means that some additional system is in place to monitor the key-generation device physically. Another idea is the implementation of a fuse designed to trip if the power level of the communication system exceeds a certain threshold, e.g. if there were some bonus high-powered laser from an unknown source heating things up.
For now, the Waterloo quantum hackers are satisfied having demonstrated a "realistic if time-consuming" method of busting a would-be perfect encryption method.