Quasi-True Random
in [Cryptography]
|
From: Phoenix on 10 Jan 2010 19:04 I have two machines with two different processor speed. First machine is faster then the second one. In booth I run an algorithm to generate random numbers with the same initial seed and clock value. The pseudo code for the algorithm is: START THE CLOCK WITH ZERO GIVE A SEED DO MAKE RANDOM NUMBER IF CONDITION THEN GET A NEW SEED FROM THE CLOCK LOOP UNTIL 1000000 As the first machine is faster, the algorithm will re-seed sooner then the second machine and the next numbers generated will be different from the second machine. My question is: Can we consider this a kind of quasi-true random?
From: Gordon Burditt on 10 Jan 2010 22:50 >I have two machines with two different processor speed. > >First machine is faster then the second one. > >In booth I run an algorithm to generate random numbers with the same >initial seed and clock value. You are generating pseudo-random numbers, not random numbers. That's an important distinction. You also aren't using a key. Using a clock as a seed, particularly one that counts seconds, doesn't leave much randomness in the seed. >The pseudo code for the algorithm is: > >START THE CLOCK WITH ZERO >GIVE A SEED >DO > MAKE RANDOM NUMBER > IF CONDITION THEN GET A NEW SEED FROM THE CLOCK >LOOP UNTIL 1000000 Loop until *WHAT* is 1000000? The clock? 20 bits of seed is pretty crappy seeding. What is "CONDITION"? Does it relate to the real world (e.g. DJIA changes by more than 20 points, up or down) or only to the algorithm (random number generated is 42)? It is very easy to cripple a very good pseudo-random number generator with a poor method of generating a seed. >As the first machine is faster, the algorithm will re-seed sooner then >the second machine and the next numbers generated will be different >from the second machine. That's assuming you don't store the numbers for later use. If you don't, you've got pseudo-random numbers for a pseudo-stream pseudo-cipher which nobody, even the intended recipient, can decrypt. If you do store the numbers, one computer piles up numbers faster than the other one, and they may or may not still be in sync. >My question is: > >Can we consider this a kind of quasi-true random? No, it's not true random numbers, it's pseudo-random numbers. You don't have any random input that I can see.
From: Phoenix on 11 Jan 2010 09:10 Loop until *WHAT* is 1000000? Loop until *1000000 pseudo random numbers* > What is "CONDITION"? Condition is: If the pseudo random number generated is > 0.999 ... The seed is a key independent from the clock. Better write, the pseudo code is: GIVE A SEED (KEY) START THE CLOCK WITH ZERO DO MAKE A PSEUDO RANDOM NUMBER (0,1) IF PSEUDO RANDOM NUMBER > 0.999 THEN GET A NEW SEED FROM THE CLOCK LOOP UNTIL 1000000 PSEUDO RANDOM NUMBERS
From: Joseph Ashwood on 11 Jan 2010 09:40 "Phoenix" <ribeiroalvo(a)gmail.com> wrote in message news:76ab2817-c343-4ead-a893-bbaf695f01b6(a)21g2000yqj.googlegroups.com... > I have two machines with two different processor speed. > > First machine is faster then the second one. > > In booth I run an algorithm to generate random numbers with the same > initial seed and clock value. > Can we consider this a kind of quasi-true random? No you can't. The difference between the clock rates will remain virtually identical, so the ratio of the rates will remain constant and it is a relatively simple matter to observe that there is no entropy introduction (technically there is a very small introduction from the jitters in the crystal, but these are functionally non-existent). Joe
From: Jens Stuckelberger on 11 Jan 2010 10:07
On Sun, 10 Jan 2010 16:04:29 -0800, Phoenix wrote: > Can we consider this a kind of quasi-true random? Quasi-true random is defined - how? |