Re: Link padding and the intersection attack

[Andrei Serjantov <aas23@hermes.cam.ac.uk>]

<busy people: this email contains nothing new! just a comment about Wei
Dai's document>

I'll take up Roger's challenge and argue a little about link padding. This
does not mean that I don't agree with Paul's email about new threat
models. That one will take longer to reply to and maybe I'll leave the
discussion of that till we meet up.

> > http://www.eskimo.com/~weidai/freedom-attacks.txt

Interesting. I had not seen that before.

The first attack works against padding to a constant. I also came up with
it (independently) when we were thinking about Tarzan back in March 2002.
There is also a protection against it -- check the archives of this
list.

Basic idea: make the maximum bandwidth of the real traffic hidden and
slightly (randomly) less than the bandwidth of the link visible to the
observer. That protects against the attack.

The second attack is the tagging attack is described well in the Mixminion
paper. (See mixminion archives).

A.

------------------
Andrei Serjantov
Queens' College
Cambridge CB3 9ET
http://www.cl.cam.ac.uk/~aas23/