On Apr 10, 2010, at 6:11 PM, Guilhem Poirier wrote:

>
> Hi
>
> I'm a french student in Master of Mathematics, computers sciences and
> application to cryptology in Paris 7 university.
>
> As part of my degree course, I have to understand and to describe the
> method of cryptanalysis of GSM encrypted communication.
>
> So I have some question about table parameters, I don't understand how
> we get the formula
>
> (1 - h*l^2 / Z)^k = (1 - c)

'c' is the probability that the next chain collides with a previous h  
chains of length l*k.

The probability that the first segment of that new chain (with length  
l) collides with any segments of the other chains is:
h * l^2 / Z   where 1/Z is the collision probability of two A5/1  
outputs.

Deriving the collision probability over the whole length of the chain  
gives the above formula.

> What is the value of Z ? In the Excel sheet
> (http://reflextor.com/Rainbow.Tables.xls) we have Z = 2/N, but in the
> case where k=1 and c=50%, it contradict the matrix stopping rule of
> Hellman's matrix because we have h*l^2 = N/4 and not h*l^2=N.

Z is what other papers call 'N'.
The matrix stopping rule randomly picks one possible design point. We  
chose another one (or rather keep it flexible through the 'c'  
parameter).

> I would like to know if it's possible to have more precision about
> stopping rule in rainbow tables.
>
> (sorry for my bad english)
>
>
> _______________________________________________
> A51 mailing list
> [email protected]
> http://lists.lists.reflextor.com/cgi-bin/mailman/listinfo/a51

_______________________________________________
A51 mailing list
[email protected]
http://lists.lists.reflextor.com/cgi-bin/mailman/listinfo/a51

Reply via email to