On Thu, Jun 15, 2006 at 09:33:12 -0400,
Brendan Dolan-Gavitt <[EMAIL PROTECTED]> wrote:
> This is covered in the Tor FAQ:
> http://wiki.noreply.org/noreply/TheOnionRouter/TorFAQ#head-a79d22244cc04ca5472832cbcc315198b875f34c
>
> The best attack that I know of right know involves measuring latency to each
> Tor node and correlating that with transmissions at the destination server.
> The latency goes up on those nodes carrying the traffic to the destination
> server when that server is transmitting data, allowing the attacker to
> determine the path through Tor (though not the original source of the
> traffic). See "Low-Cost Traffic Analysis of Tor" for more details:
There can be other attacks in special cases. Tor users are rare and if you
know some other information about them (like that they are a user on your
network) then it might not be too hard to figure out who they are.
I used this idea a long time ago to figure out who sent an anonymous threat
to one of our users through anon.penet.fi remailer. They were the only
user on our system to have sent email to that remailer at a time close to
when the threat message was received.
This sounds basically like an intersection attack--you take the set of anonymous users active at the times you're interested in, and intersect them to find out which user it was. Intersection attacks are still an open problem for every anonymous network that I know of, but they still require some outside knowledge that isn't usually available to an attacker--namely, the set of people connecting to Tor at any given time. If you run a corrupt Tor node and log the IPs of people who connect to you, there's a good chance that eventually you'll get the information you want--but as the Tor network grows, your chances get worse and worse.
-Brendan
_______________________________________________ Full-Disclosure - We believe in it. Charter: http://lists.grok.org.uk/full-disclosure-charter.html Hosted and sponsored by Secunia - http://secunia.com/
