Physicist Uses Laser Light As Fast, True-Random Number Generator

MrKevvy writes "An Ottawa physicist is using laser light to create truly random numbers much faster than other methods do, with obvious potential benefits to cryptography: 'Sussman's Ottawa lab uses a pulse of laser light that lasts a few trillionths of a second. His team shines it at a diamond. The light goes in and comes out again, but along the way, it changes. ... It is changed because it has interacted with quantum vacuum fluctuations, the microscopic flickering of the amount of energy in a point in space. ... What happens to the light is unknown — and unknowable. Sussman's lab can measure the pulses of laser light that emerge from this mysterious transformation, and the measurements are random in a way that nothing in our ordinary surroundings is. Those measurements are his random numbers.'"

You should have said

[MurukeshM]

Obligatory xkcd: http://xkcd.com/221/ [xkcd.com]

Skip the newspaper article...

[Vario]

The newspaper article is not giving any information that is not already included in the summary.

The paper is published in Optics Express, the abstract can be read here [opticsinfobase.org]. The full article is behind a paywall unfortunately. The author claim that this concept could deliver random numbers at a rate of 100 GHz which is quite fast compared to other true random number generators out there that are based on thermal noise, radiation or other processes.

Re:You should have said

[Anonymous Coward]

Sigh. You kids who can't remember 10 years ago [kodyaz.com].

Re:Finally a reason for socially inept people to b

[Hentes]

A lot of tools have diamond blades.

Re:"Truly random numbers"

[medv4380]

Incorrectly applying the conservation of information. What you are saying wouldn't work and would violate the Heisenberg uncertainty principle. What would happen is as you did the measurements to that degree you would lose information on the motion of the particles involved as you gained the new information on the position hence the conservation of information. The uncertainty of the choices would still exist and you'd most likely get two different results if his choice had any quantum affects involved.

Re:"Truly random numbers"

[Baloroth]

The term "random" is generally (even in science, from what I know of it) taken to refer to things which we are not able to predict, even theoretically. We do not, however, know for sure if the system is non-deterministic (that is, truly random) or only apparently so.

Again, not a quantum physicist. But I believe that is the general state of affairs. See Wikipedia [wikipedia.org] for more.

Re:Simtec "Entropy Key" also does quantum RNG

[evanbd]

You can also use resistor noise [wikipedia.org], a good amplifier, and an ADC to make moderately high bandwidth true quantum RNG. I priced out a simple design with a microcontroller on a USB key footprint; looked like $50-100 in prototype quantities, less in large quantities, for 10 KB/s output (or so). Getting the entropy is looked like the easy part; it then needed a fair bit of CPU power (by microcontroller standards) to hash that into usable bits.

You can also (with a lot more software work, and low bitrates) use the resistor noise present in audio input channels to good effect. Turbid [av8n.com] is a project that does just that. Note that when evaluating such projects, the hard part is not getting the numbers, but proving that they have enough entropy, and that they've been properly processed to preserve it. Turbid does an excellent job on this important documentation step.

Re:Finally a reason for socially inept people to b

[kelemvor4]

Oh yeah baby... back with my VLB Diamond Viper 4MB of VRAM... and a 486DX-2 66 with 16MB I was styling'. The chicks just couldn't stay away.

You must have been rolling in the dough back then to have 16MB ram. 8MB about broke my bank.

Re:Finally a reason for socially inept people to b

[CODiNE]

Yeah I was a spoiled brat.

More obligatory Dilbert

[artor3]

Dilbert did it first, and better.

http://dilbert.com/strips/comic/2001-10-25/ [dilbert.com]

Re:What's the distribution?

[martas]

The full paper (link by "Vario" above) seems to indicate that they get random bits, i.e. coin tosses, and claim that they can get a uniform distribution (i.e. 0.5). They also say that "Any possible bias in the phase measurement is removed by post-processing using a fair bit extractor algorithm", citing two papers* (i.e. that though their measurements could lead to a slightly different distribution, they can correct for that). I'm not familiar with the technique, but I guess it's well established. They also show results and say that they did something called the "DIEHARD statistical test suite" (which is apparently a set of tests designed exactly for this problem, i.e. random number generation), and "confirm[ed] that the measured optical phase is a suitable source of random numbers", though I'll have to take them at their word because I'm not familiar with the theory behind this.

* First:
J. von Neumann, “Various techniques used in connection with random digits,” Nat. Bur. Stand., Appl. Math Ser. 12, 36–38 (1951).
Second:
A. Juels, M. Jakobsson, E. Shriver, and B. Hillyer, “How to turn loaded dice into fair coins,” IEEE Trans. Inf. Theory 46, 911 –921 (2000).)

Re:C64

[Anonymous Coward]

The noise produced by the sid chip is from a linear feedback shift register, so is not really random.

Re:"Truly random numbers"

[bhagwad]

Just because we can't know a position and vector of a particle now doesn't mean we won't ever be able to.

Yes it does. The Heisenburg uncertainty principle is not a limitation of technology. It's a law of the universe. It's like saying "just because gravity exists today doesn't mean it will exist sometime in the future."