In April of this year I wrote about how quantum cryptography (more properly called Quantum Key Distribution or QKD) was leaving the laboratory bench and is balanced on the cusp of entering into real-world use. At the time, many thought I was talking about the far distant future. However, as is so often the case, people underestimate how rapidly technology moves on. A paper just published by a research group in Cambridge UK could be a game changer for QKD.
One of the main objections to QKD has been the expense. It has been necessary to have fibre optic cabling dedicated to the task (so called dark fibre). With dark fibre, in the absence of other data signals, secure key rates exceeding 1 Mb/s and a transmission distance of over 250 km have been achieved. Until now, the idea of leasing such fibres from telecoms providers has put potential users off at the first hurdle.
However, the results of the Cambridge team provide a technique where QKD can be used on a fibre optic cable that is already being used for other communications traffic. The ability to use shared fibres has suddenly, and to many people’s surprise, made QKD an economic prospect, with costs likely to be little different to a corporation fitting a top-end firewall.
QKD and traditional data traffic are at opposite ends of the brightness scale. When beaming traditional data down a fibre, you try to use high power levels in order that it will travel as far as possible without the need to “repeat” the signal. QKD, however, requires light intensities at the lowest imaginable levels. With QKD, we are trying to send and detect single photons of light. Not only that, but we need to be sure that our transmission method isn’t interfering with the “quantum state” of the photon, which is the very property upon which QKD relies.
I describe the task of using shared fibres for QKD as like trying to see the stars whilst staring at the sun. One simply overwhelms the other.
What the Cambridge team have succeeded in doing is switching between the various light sources so rapidly, and so cleanly, that very small numbers of photons can be sent and detected in between the pulses carrying the traditional data. The timing needs to be extraordinarily accurate as the system opens a “gate” to let through single photons for only a tenth of a billionth of a second.
The trick is that the pulse that contains the QKD photons has to be coordinated with the detectors, so that they know when to expect the one QKD photon between the pulses of millions of standard data photons. By using such accurate “time slicing”, it is now possible to at one instant stare at the sun and then momentarily see the stars.
The technique announced is so accurate that the speed at which QKD can be transmitted is not materially different to what was achieved using dark fibre, whilst the rates for standard data can be maintained at Gigabits per second. Currently the method has been used successfully on fibres up to 90km long, but just as distances for QKD over dark fibres have increased up to 250km, it is highly likely that these distances will be extended. After all, this is just the first announcement of the technique, and the one thing we know for sure is that technology has a habit of getting better very quickly.
Of course, nothing is ever quite as simple as it appears. Fibre optic cable suffers from effects that can work against this time slicing technique, no matter how accurate it is. One such phenomenon is called Raman Scattering. This is different from the more usual Rayleigh Scattering that people tend to envisage with fibre optic cabling, as they imagine the photons bouncing their way down the length of a glass fibre. In Raman Scattering, rather than bouncing inside the glass crystal of the fibre “elastically” so that the photons maintain their energy and wavelength, a very small proportion of the photons (1 in 10million) are scattered so that their frequency is changed. This causes detection problems.
However, the phenomenon is well understood, the models predicted to a high degree what was observed by the Cambridge team, and their system coped well, and was able to do the necessary error correction. Indeed, the fact that the Raman scattering modelling was so well modelled in these tests suggests that the team have an environment that they can control successfully. If you can predict the level of errors likely to occur, then you can make the necessary corrections. You treat the errors as noise, which in the field of communications engineering is a well-trodden path.
More prosaic matters are what will hold up this technology. For domestic use, there are few homes that have fibre running all the way to the homes router. But, certainly in the UK, fibre is making an appearance at the end of many streets in cities across the nation, even though the final leg of the network is good old fashioned copper. Governments are vying to roll out superfast broadband in their countries, which can only mean that fibre based networks will proliferate. Perhaps with them will come “super secure” networking in the form of QKD.
Meanwhile, fibre networks are the basis for most Internet backbone providers, and many corporations. With the costs now of the same order as much of the other equipment they buy for securing their networks, we must surely be about to see QKD become at least part of the network security landscape.
It appears the Cambridge research team have made a significant step forward, and whilst the journey is far from complete, I believe this further demonstrates that QKD is no longer just of academic interest. It’s about to become a mainstream method of securing data networks.
Image: Tommy Moorman, from Best-Kept Secrets by Gary Stix, Scientific American January 2005 Issue.