A team of researchers from University of Toronto Engineering has discovered hidden multi-dimensional side channels in existing quantum communication protocols.
The new side channels arise in quantum sources, which are the devices that generate the quantum particles — typically photons — used to send secure messages. The finding could have important implications for quantum security.
"What makes quantum communication more secure than classical communication is that it makes use of a property of quantum mechanics known as conjugate states," says PhD student Amita Gnanapandithan, lead author on a paper published in Physical Review Letters .
"For example, position and momentum are conjugate variables: when you measure one, you disturb the other. If both variables are randomly chosen for encoding, anyone trying to listen in on the message will automatically introduce disturbances that can be detected by the parties that are trying to communicate.
"Furthermore, due to the quantum no-cloning theorem, the eavesdropper cannot produce copies of the message to listen in on."
Yet despite its inherent security, there are still some ways that quantum communication can be compromised due to the imperfections in the devices used for practical implementations.
Between 2000 and 2012, researchers showed that side channels can arise because of the way that quantum detectors work. These side channels act as loopholes, enabling someone to listen in on the signal without introducing a detectable disturbance.
To address this, in 2012 Professor Hoi-Kwong Lo and his collaborators developed a new protocol known as measurement-device-independent quantum key distribution (MDI-QKD) . The protocol effectively short-circuits all side channels associated with quantum particle detectors.
With the detector taken care of, Gnanapandithan, who is co-supervised by Lo and Professor Li Qian, turned to looking for potential side channels associated with the other end of the communication: the source devices.
"Let's say you want to encode information based on how the light coming from the source is polarized, which we call optical polarization," says Gnanapandithan.
"You would use two conjugate polarization bases to perform encoding and ideally, you'd want to keep your encoding within the polarization degree of freedom. You also don't want that polarization to be correlated with any other degree of freedom, because if it is, then the eavesdropper can measure that second one to get information on polarization."
The idea that the encoding degree of freedom is uncorrelated with other degrees of freedom in optical quantum sources is known as the dimensional assumption. Violating this assumption means that the message might not be secure.
In practice, today's quantum sources can often introduce such a violation due to, for example, correlations between adjacent signals. This is called the pattern effect, and results in information about earlier signals leaking into later signals.
But in the most recent study, Gnanapandithan used both theoretical models and physical quantum sources to demonstrate a new source of the violation that hadn't been considered before.
"We knew that the modulation process can be a little bit distorted, but what we found was that the modulation process can also be time varying, even within the same signal optical pulse," says Gnanapandithan.
"Specifically, we made the very subtle realization that this flaw is actually a violation of the dimensional assumption. Hence, we call this type of flaw 'hidden multi-dimensional modulation,' of which time-varying encoding is only one example."
How big a problem these side channels are depends on the type of equipment being used.
"If your equipment has a higher bandwidth, you can apply a modulation signal to your optical pulse that should get it closer to what it ideally should be," she says.
"But if your equipment is severely bandwidth limited, then the modulation pulse might be severely distorted, and that would worsen the issue.
"There is also a new type of quantum key distribution (QKD) source that's been introduced in the literature, called a passive QKD source. Passive QKD sources don't even use modulators, so these bandwidth issues wouldn't apply."
Lo says that future work from his team will focus on possible ways to mitigate the newly discovered side channels.
"We can get creative, and perhaps find ways around these problems," he says.
"But as we've learned in the past, it's also possible that our new method might give rise to its own problems. You never know how many layers there are going to be, but I think the all-important first step is to simply identify the issues you have to deal with, and that's what we've done here."
