[tor-talk] Using the middle relay to guard against correlation attacks.

[benjamin.wheeler at vfemail.net]

First off I'm not a computer scientist. Nor am I a Tor expert.

I'm wondering if it is possible to use the middle relay as a buffer to  
protect against possible correlation attacks.

 From my understanding, if the attacker controls the first relay, and  
the last relay, she can transmit packets at a certain burst rate, and  
size to generate a visible pattern that can be detected at the other  
end.

Unless the middle relay interferes in reshaping that pattern.

So what if in creating the circuit, the client would ask the middle  
relay to buffer the traffic at a certain buffer size and at a certain  
timer variable? The timer variable is used in the case the buffer does  
not fill up.

So when the middle relay receives incoming or outgoing traffic for  
that circuit, it would buffer the data until the buffer is full, then  
transmit, or until the timeout of the timer has elapsed since first  
bits of data started to buffer then transmit.

We make the client request from the middle relay to allocate the  
buffer size and timer in milliseconds, and if they are both 0, then  
the relay behaves as normal as it is currently.

We also can make the relay set it's own parameters on what the max  
buffer size should be and max timer variable allowed. If the client  
circuit creator is asking for too much, or inconsistent values, either  
give the defined relay max, or ignore, or send back an error to the  
client.

Is something like that possible to implement in Tor? Does TCP allow  
it? How well will it scale?


-------------------------------------------------

VFEmail.net - http://www.vfemail.net
ONLY AT VFEmail! - Use our Metadata Mitigator to keep your email out of the NSA's hands!
$24.95 ONETIME Lifetime accounts with Privacy Features!  
15GB disk! No bandwidth quotas!
Commercial and Bulk Mail Options!