Quantum random number generator

Darkchilde · #1 by **Darkchilde** » Nov 30, 2011 1:00 pm

From here: http://www.physorg.com/news/2011-11-einstein-dice-quantum-random.html

(PhysOrg.com) -- Quantum mechanics implies that uncertainty in experimental measurements are an inherent part of nature – an idea that Albert Einstein disparagingly characterized as “rolling dice”. This true quantum randomness, for which Einstein was concerned, contrasts with a conventional gaming die, whose motion follows the laws of classical mechanics and is therefore pseudo-random. With the right physical information about initial conditions, the outcome of a dice roll can be accurately predicted.

Now, reporting in the online issue of Optics Express, a National Research Council (NRC) team led by Dr. Benjamin Sussman has successfully used quantum mechanical fluctuations to create a physical analogue of truly a random die. More importantly, their die can be rolled extremely quickly and can be easily measured providing the potential to transform the security of future high-speed information networks – from encrypting military communications, to securing individual online purchases, to generating random numbers for lotteries, or in high performance computing applications.

Continued here: http://www.physorg.com/news/2011-11-einstein-dice-quantum-random.html

The relevant paper can be found here: http://www.opticsinfobase.org/oe/abstract.cfm?URI=oe-19-25-25173

Very interesting.

The_Metatron · #2 by **The_Metatron** » Nov 30, 2011 3:46 pm

I don't get this. It's been trivial to produce truly random numbers by sampling the noise from a reverse biased diode for years and years. Which is produced by quantum tunneling. So what?

It's not that terribly useful in encryption because it's not repeatable. You can mix a signal with a true random number and encrypt it. Easy peasy. An AND gate will do it. But the problem is getting that identical random number to the decryptor, and doing it securely.

It's essentially the same as what's called a one time pad. Very, very, secure. But insecure at the same time, because of the vulnerabilities inherent in transmitting the key itself.

twistor59 · #3 by **twistor59** » Dec 01, 2011 7:48 am

I remember when I was a kid, reading in Practical Electronics magazine, that you produced random numbers by sampling the noise in a reverse biased zener diode. Very cheap and easy, and not pseudorandom.

natselrox · #4 by **natselrox** » Dec 01, 2011 7:53 am

What is the technical difference between a PRNG and a True RNG? How do we know it? Can anyone please explain it in terms a layman can understand?

The_Metatron · #5 by **The_Metatron** » Dec 02, 2011 2:46 pm

Sure. A pseudo random number only appears random. But, it can be recreated. Useful in cryptography where you mix your plain text with the crypto key (either the pseudo random number, or something derived from one). The resulting encrypted text will appear to be random digits.

But, if your friend can reproduce that same pseudo random number you used to encrypt your secret, he can use it to decrypt the cypher text to arrive at the same plain text message.

True random noise is not repeatable. Great for encrypting, but unless you recorded the random string used to encrypt, your message is lost forever. No way to reproduce the random string used to encrypt it.

How to tell a pseudo random number from a true random number? Fucked if I know. That's the whole point of encryption. To make a pseudo random number that is indistinguishable from real random noise.

Cryptanalysis fucks with my head.

susu.exp · #6 by **susu.exp** » Dec 02, 2011 4:29 pm

Well, PRNGs create binary strings that look like binary strings produced by a RNG up to some point, where the latter puts out 1s and 0s with a probability of 0.5 and all digits are independent from one another. There´s some number N, so that afterwards the PRNG doesn´t do this - often there´s periodicity for instance, the string repeats. in a lot of applications pseudo-random numbers are used, because they have the advantage of being repeatable given the same seed and you can make N rather large (large enough for the specific application).