Better security achieved with randomly generating biological encryption keys
phys.org
phys.org
Far as dice, one of my experiments in TRNG's was a shoe box full of dice. Shake it up, dump them, put the numbers into a computer program, whiten it, and tada. The first one used 8-sided die since each one maps cleanly to 3 bits (2 ^ 3 = 8). The next iteraiton used regular, 6-sided die just discarding any above four to map easily to 2 bits. Did something similar with a deck of cards after throughout shuffling.
Such methods were good for producing an initial secret that I'd use with algorithms like CPRNG's to produce the rest.
I was under the impression that shuffled card deck is more difficult to convert to random number, you need something like Lehmer Code. Did you find some easy good shortcut here?
If you shuffle a deck of 32 cards, you get ~117 bits, a fair bit less that the 160 bits you would get for 5 bits/card, and there's extra math.
1. 32 = 117 bits of entropy, not 150, after lots of shuffling.
2. It works if I randomly select a single card at a time but not in order from a randomized deck. In order from a randomized deck should give the entropy amount or close to it.
I'm not a probability expert. Just doing what I could with good, shuffling skills. Good enough that people refused to play Spades with decks I shuffled. So, I thought I had good entropy. I'm always up for improvements.
The second card can be any one of 31 cards, because it can't be the same as the first card. This gives you log2 31 = ~4.95 bits of entropy.
If this argument isn’t convincing, think of it this way. If you draw two cards, then there are 32 choose 2 cards you can draw, and 32 choose 2 is 496, giving log2(496) = ~8.95 bits. You can draw them in two different orders, giving 1 extra bit, total of ~9.95. If you could draw a pair of the same card, there are 32 different pairs to draw, giving (496 * 2) + 32 possibilities, which is 1024, and log2 1024 = 10. But you can't draw a pair of the same card.
Or for another argument, consider that there are 32! different ways to shuffle the deck with 32 cards. log2 32! = ~117.
It's great to do research like this as that is how we get advancements and breakthroughs, but it's another thing to completely ignore an entire sub-field of the area you're researching in order to misrepresent what you're doing as being an amazing breakthrough.
All it would have taken is a simple statement like "We acknowledge there are existing non-pseudo-random solutions to this problem that use thermal noise, radioactive decay, etc, but wanted to try a biological process because ____________".
Bit of a tangent but do you have any particular devices in mind? I did a cursory search, and found only Microchip (nee Atmel) ATSHA204A and the related family of products, but those are full-blown fat crypto processors. Are there some cheaper/simpler plain TRNG chips available somewhere?
> Alice and Bob, secretly, exchange their private key, X, generated using the spatiotemporal randomness of the T cell population.
It's always interesting to read about new RNG methods, but that seems pretty far from a crypto system.