back to article Boffins: How to generate crypto-keys using a smartphone – and quantum physics

Your smartphone can be used to generate cryptographic keys from truly random numbers "of a quantum origin", according to bods at the University of Geneva. The Swiss research claims, quite simply, that illuminating the camera of a device like the Nokia N9 can cause quantum effects, which ultimately can be used to generate …

COMMENTS

This topic is closed for new posts.
Silver badge

Just wondering

" ... the right number of photons needed to balance getting the maximum quantum uncertainty, while not saturating the detectors."

That indicated they were using a bright 'scene', if they were worried about saturating the CMOS detectors. I'd have thought that quantum effects wouyld be more noticable at very low levels of illumination.

Has anyone tried generating a 'random' number my performing a checksum on all the pixels of a photo of a natural scene (flower beds in a park, woodland glade, etc)?

2
0

Re: Just wondering

Good idea. I might experiment with that checksumming photos technique and see what I come up with.

I'll hold your post up as prior art should some innovation-stifling patent troll try to patent this. Would you be generous enough to put your idea into the public domain for all to use? If I invent something worth marketing I'll slip some of the proceeds your way!

2
0
Silver badge
Happy

Re: Just wondering

I hereby relinquish all rights and claims to the random number generator method known as the 'Frank Ly Generator' and gift it to the world. (I retain the right to bask in any reflected glory.)

Note: I had to press a 'Sumbit' button to get the post published. Vulture lawyers might have a say in this.

4
0
Silver badge

Re: Just wondering

It probably is a good idea until some shady folk come along and sit on the (fictional for now) world development panel which then decides that the world should be covered in concrete with one colour and a smooth finish and the sun fixed in one position in the sky. C

In unrelated news cameras will be fixed to be unable to take pictures at night (for the sake of the children.)

Cheers

Jon

2
0
Anonymous Coward

Re: Just wondering

Yes, they have generated random numbers from photos. They used Lava Lamps. Patented by SGI, so don't try this at home kids.

https://en.wikipedia.org/wiki/Lavarand

3
0
Anonymous Coward

Re: Just wondering

"It probably is a good idea until some shady folk come along and sit on the (fictional for now) world development panel which then decides that the world should be covered in concrete with one colour and a smooth finish and the sun fixed in one position in the sky. C"

You can get good entropy from a webcam even if the lens is covered up with opaque tape. It's the inevitable noise at the low-order bits that does it, so the actual image is irrelevant.

1
0
Silver badge

Re: Just wondering

No idea if it's better/worse, but it seems simpler, as you said, to just get a number from a random sample of an image taken with the phone and immediately delete (to stop/prevent recovery of the photo to generate a key again).

Though a single hardware quantum input is easier than asking a customer not to take a photo of the inside of their pocket, and choose something truly "unique" instead.

0
0

Re: Just wondering

Out of patent in less than two years.

0
0
Silver badge

It may be random but is it unbiased?

How do they manage to make it indistinguishable from perfectly random? Even if it has a quantum input, the process of digitisation might make it more likely for instance to get a '1' than a '9' - the article doesn't say how they get round that.

Would they be able to make certain that no production devices had any such bias?

2
0
Silver badge

Re: It may be random but is it unbiased?

Most random number generators that use real-world inputs tend to apply a little "bleaching" or "whitening" to the raw input to help trim out any unintentional biases. It's also possible to run the inputs through assorted randomness tests to see if the stuff passes muster.

1
0

Sorry Frank - already exists

www.ijesit.com/Volume%202/Issue%206/IJESIT201306_68.pdf

at least once

ISSN: 2319-5967

ISO 9001:2008 Certified

International Journal of Engineering Science and Innovative Technology (IJESIT)

Volume 2, Issue 6, November 2013

544

A New Randomized Cryptographic Key Generation Using Image Priyanka.M, Lalitha Kumari.R, Lizyflorance.C and John Singh. K School of Information Technology and Engineering, VIT University, Vellore, Tamil Nadu, India

1
0
Go

Awesome!!

I so want to use this, but http://xkcd.com/221/

3
0
Anonymous Coward

Sounds a lot like the video noise entropy technique, which I've dabbled with in Linux. Basically, the low-order bits of the pixels of a webcam picture are too subject to noise to be that reliable for the picture; but then that makes it a potential source of noise. And you don't even need an actual subject; it works even with the lens covered in duck tape. Depends on the camera, of course, but based on my research, I've found that most basic webcams are noisy enough to produce data that, bleached a couple times, can pass the FIPS-180-2 randomness tests and still produce them at a decent clip (I had a VGA webcam produce a MB of randomness in a couple minutes).

1
0

And how exactly does one lock down a phone these days?

The one device I trust least to hold sensitive data is my cellphone. What am I meant to do if it's my key generator? Never put a SIM into it?

1
0
Gel

Bunkum: digital noise will adversely affect the randomeness

Electrical noise from the phone will have a significant effect on the randomness, and this should be easy to detect. An example is to turn the volume up on a silent audio track. That audible noise in theory is random and quantum in nature, but because of the digital noise generated by the phone actually gives a characteristic digital noise sound. It is quite hard to eliminate the digital noise and get truly random quantum noise. That is partly why quantum generators, while simple in principle, are actually quite expensive.

Even when the digital noise is inaudible it still affects randomness and can be detected. There is unavoidable digital noise generated within the image sensor and it is not a good choice for a quantum random number generator. Finding a perfectly steady light source with only quantum noise is also a challenge.

0
0
Silver badge

Re: Bunkum: digital noise will adversely affect the randomeness

Actually, some random number generators RELY on that digital noise to produce entropy. These systems, however, recognize the potential for patterns which is why they tend to bleach the raw data to get rid of those biases. I've done this very thing using webcams, and while not all webcams are created equal, a couple different bleaching passes usually produces quite acceptable results.

1
0

10^18 times the life of the universe to crack?

More fool them. I have TWO computers working on it, so I can cut that time in half.

2
0
Silver badge

Re: 10^18 times the life of the universe to crack?

Only thing is that half of a big fat number is still a big fat number: in this case, 5 * 10^17. It's like with those big lotteries. Sure, you slash the odds with each additional ticket, but the odds are still pretty crazy. And there's no easy way to reduce the all-important exponent to the extent that it becomes feasible within our lifetime.

0
1
This topic is closed for new posts.

Forums

Biting the hand that feeds IT © 1998–2017