Re: More Mix questions

["Michael J. Freedman" <mfreed@cs.nyu.edu>]

On Sun, 11 Jan 2004, [iso-8859-1] Some Guy wrote:

> Date: Sun, 11 Jan 2004 19:02:31 +0100 (CET)
> From: "[iso-8859-1] Some Guy" <amichrisde@yahoo.de>
> Reply-To: or-dev@freehaven.net
> To: or-dev@freehaven.net
> Subject: More Mix questions
> 
> I was wondering: has anyone developed circuits that branch out in a tree?  It just seems like a
> cool way to solve this problem.
> 
> As with SURBs, has anyone considered single use tunnels?  They seem like a good idea for a network
> where nodes only know about thier neighbors.  It might still be economical for large operations.

Both of these (at least, as far as I can understand what a tree-based
structure might look like), would actually compound the problem caused by
dynamic routes.  See, for instance, [1] and [2]:  Namely, there is some if
you can link route 1 and route 2 as part of the same session, then the
initiator is your immediate predecessor with higher probability than any
other node in the network.  To gain confidence as to who is the initiator,
you need a bunch of such routes; the more dynamic your paths are, the
quicker to learn that critical number of routes.  (See [2] for full
explanation.)

--mike


[1] Crowds: Anonymity for Web Transactions (PDF, gzipped PS) (Cached: PDF,
gzipped PS) by Michael Reiter and Aviel Rubin. In ACM Transactions on
Information and System Security 1(1), June 1998.  (BibTeX entry)

[2] An Analysis of the Degradation of Anonymous Protocols (PS) (Cached:  
PS) by Matthew Wright, Micah Adler, Brian Neil Levine, and Clay Shields.
In the Proceedings of the Network and Distributed Security Symposium -
NDSS '02, February 2002. (BibTeX entry)

See http://www.freehaven.net/anonbib/date.html


-----
"Not all those who wander are lost."           www.michaelfreedman.org