> A question: assuming, you have a class of random number generators with > lots of internal state. (Lots: like >>10^6 bits). Let's say the evolution > through state space of that generator is provably reversible (or nearly > reversible), and that the Hamiltonian of the system is stochastic (system > evolution is a randomwalk in state space). The result is a pseudorandom > number generator with a ridiculously long periode, and good randomness of > output, obviously. A simple cypher based on it would exchange the > pseudorandom generator state (the key) through a secure channel, > similiarly to a one time pad.
I think that the patent implies true hardware rng. As for PRNGs, if you can exchange million bits securely, the desired unicity distance (based on your paranoia level) will determine how often you must re-key. I am not sure that there is a *simple* prng with 10^6 bit state. Feeding million bits to 1-DES 64 bits at a time and using the output as the key for the next cycle could be one way for dilluting entropy. ===== end (of original message) Y-a*h*o-o (yes, they scan for this) spam follows: Yahoo! Greetings - Send FREE e-cards for every occasion! http://greetings.yahoo.com
