On Mon, Dec 10, 2007 at 12:07:02PM +0100, Heiko Gro? wrote:
> Hi there,
>
> During the last weeks I've done research. I'm looking for a topic for my
> diploma which should be close to anomity/tor. I've read quit a few about
> cicuit-building, geoip and node-exclusion in order to build
> "Vorratsdatenspeicherungs"-proof circuits. At the beginning of my
> research I thought it would be suitable to just restrict the choice of
> the nodes by writing a special controller, but later on I became aware
> of other possible kinds of attacks such as statistic attacks on both
> ends of the communication.
> Where are you guys right now at? Is there anything I can do?
> I'm looking forward to find some working specially at this kind of problem.
The state of the art on resisting end-to-end statistical attacks by an
attacker who can watch both ends of a low-latency anonymity channel,
is IMO Shmatikov and Wang's "Adaptive Padding":
http://freehaven.net/anonbib/author.html#ShWa-Timing06
The paper mentiones earlier research in the area, which is also pretty
important. IMO none of it is useful enough yet to merit inclusion in
Tor, but it might be on the right path.
For a paper with interesting implications for path selection stuff, I
like Steven Murdoch's
http://freehaven.net/anonbib/author.html#murdoch-pet2007
It refers to some earlier papers that are also pretty important.
You might also check out other related papers on anonbib; there's some
good research been happening in the last few years. IMO, new defenses
are generally more interesting than new attacks, since the attacks are
already so good that new, equally good attacks aren't very impressive.
hope this helps,
--
Nick MathewsonAttachment:
pgpROikTaNgsO.pgp
Description: PGP signature