Author: David WagnerDavid Wagner
Date: Dec 2, 2006 11:28
David Eather wrote:
>if in the standard DH key exchange some sod called Mallory manages to be
>man in the middle, does his attempt fail if the two uses have an
>advanced type of secure token?
>
>The bank sends Bob a challenge made up from their DH key with Bob (or
>Mallory) and a secret key function from a secure token. Then Bob enters
>that number into his secure toke plus his DH info and sends that back to
>the bank. The bank checks and that doesn't verify either - and Mallory
>is exposed! or not? Does this idea work.
This isn't standard DH any more. You are essentially trying to invent
your own key exchange protocol. That's a risky exercise, because such
protocols are susceptible to subtle flaws that can be hard to spot.
The answer, by the way, is that I don't think it is necessarily secure.
For instance, if Mallory uses 0 as her DH exponential with both parties,
then the DH key they compute will be 0 on both sides, and the attack
will succeed. But even if you fix it so that I can't immediately find
an attack, I would still have some reluctance to use a new key-exchange
protocol without some hefty analysis to back up its security.
| 
