On 6/15/2012 2:37 AM, KZK wrote:
But Eve, who is listening in to the publicly
available noise, does not know which resistor was connected at each
end and cannot work it out either because the laws of thermodynamics
prevent the extraction of this information from this kind of signal.
So why isn't this susceptible to a simple man in the middle attack?:
Eve cuts the wire between Alice and Bob (AB line) and insert her own
node that connects to Alice (AE line) and Bob (BE Line) individually.
Alice can't tell the difference between the AB line or the AE Line and
sets her resisters. Eve sets her resisters connected on the AE line
to random and deciphers the sequence that Alice used. Eve then Uses
that sequence on the BE Line. Bob can't tell the difference between
the AB line and the BE line, sets his resisters randomly and decodes
the message. (Eve can even send Bob a False message).
Seems like this method requires a 100% secure land line, which is
impractical.
KZK--
I believe that Alice and Bob are doing the resistor thing for each bit
simultaneously,
and sharing their measurements over a separate open channel. (The paper
says the
voltage/current data on the noisy channel is "public".) Furthermore,
they're tossing
all the trials where those data show they both picked the "high"
resistors or both
picked the "low". So all Eve can usefully look at are data for
essentially identical
trials, each one with the noise characteristic of one high and one low
resistor on the
channel. Eve is free to relay noise between the two lines in your
example, but that
won't help her.
If the land line is tapped in a useful manner, the claim is that Alice
and Bob can
detect that it is. So they'd need a land line, but wouldn't have to
secure it.
---David
_______________________________________________
http://box535.bluehost.com/mailman/listinfo/brin-l_mccmedia.com
