Deprecated: The each() function is deprecated. This message will be suppressed on further calls in /home/zhenxiangba/zhenxiangba.com/public_html/phproxy-improved-master/index.php on line 456 George Danezis's Home Page
Please note this page is not activelly maintained any more. My
new web-page can be found at Microsoft
Research.
Myself
I am a post-doctoral visiting fellow at the Cosic group (Privacy
group), ESAT, K.U.Leuven, in Flanders, Belgium. I
am funded by the FWO (Flamish
research council) and K.U.Leuven to work on privacy, anonymity
and traffic analysis. My boss here is Bart Preneel.
Previously I have been a research assistant in the Security Group,
of the Computer Laboratory of
the University of Cambridge,
working on anonymous communications, peer-to-peer networks and
censorship resistance. I also got my Ph.D, M.A. and B.A at the
Cambridge University, Computer Laboratory under the supervision of Prof. Ross Anderson.
We have done some work on the Economics of
mass surveillance and the (questionable) value of anonymous
communications. We look at target selection strategies for
maximizing
surveillance (or disruption) return based on data collected from a real
social network. It turns out that current anonymous communications
solutions do not pretoect too well against such target selection. Our
data sets and scripts are available.
During my CST Part II Project I have done an experimental
implementation of Mixes and Dining cryptographers networks. The code is
available here (tar.gz)(The
whole thing is highly experimental and one should be mad to use it for
anything important). The project report explaining the implementation
issues is available (pdf). As
a result of this work I gave a talk in the Cambridge Protocols Workshop
2000 on Money Escrow, a method to combat dead-beat biders in auction
protocols. A revised version of the slides used in this talk is
available (pdf,ps).
I have been working with Richard
and Markus assesing
real life systems that provide Anonymous services. We document real
patterns of failure in these systems, and work towards a security model
for pseudonymity (pdf,html).
Some people have been sending me anonymous emails without
including a reply block, therefore not giving me the ability to answer
their questions. For this reason I have built a new page with replies to anonymous emails. Make sure
you access it using a suitable anonymizing proxy!
Anonymous communications:
What can you do with traffic analysis? Often people ask me the
same question so I have a presentation
and a background paper introducing the
topic. They were both prepared for my talk at the Santa's Crypto
Get-together in Prague, December 2005.
In order to better evaluate anonymous systems and attacks against the
anonymity properties of systems,
Andrei Serjantov
and I, propose a new definition of the "Anonymity set". We move away
from the classic world of set theory toward a definition that takes
into account probability distributions over different participants and
redefine anonymity sets using entropy and other tools borrowed from
information theory. This metric allows a better qualitative
understanding of anonymity and allows researchers to move beyond the
typical all-or-nothing approach to these systems and their failures.
The paper that describes these definitions is available in ps and pdf format. It
appeared at PET2002, and got an
award
at PET2003.
Lately I have been working on traffic Analysis, and a preliminary poster of some results appeared
at the "Workshop on Privacy and Identity in the Information Society:
Systemic Risks" (5-6th February 2002).
Along with others, we are collaborating to design and build MixMinion,
the next generation of anonymous remailers. They should support
sender-receiver anonymous communications, and support for forward
security on the links. The development lists
are public, and the first design document
is also available.
In order to strengthen mixes against new legal attack, such as
compulsion to reveal keys, I have proposed the design of forward secure mixes. Using key updating
techniques, even the intermediate nodes that have in the past processed
the message cannot trace it back, provided they follow the protocols. A
paper presenting these ideas has been presented at the NorSec2002
conference.
Other security mechanisms:
Mike Bond and I just published a new technical report entitled "A
pact with the Devil" (see Techical
Report 666).
We look at how viruses may give benefits to the owners of the computers
they propagate on in conjunction to using threats and blackmail to
entrench themselves.
I have spent some time exploring the use of javascript as a way to
implement cryptographic algorithms. A system implementing ElGamal
encryption can be found here.
We have been implementing with Richard "Chaffinch", a
system that provides confidentiality and plausible deniability using
only authentication primitives. The paper destibing it in detail can be
found here [pdf,html]. The
Chaffinch system has its own
web page.
Policy issues:
Some original sources relating the technical details of the
latest Greek interception scandals were translated into English.
In December 2001 I took part in the EU Cybercrime forum on the
subject of the retention of traffic data, as part of the Internet
Rights Europe initiative. My positions on the subject, and the other
contributions, can be found here.
Other sources of information about data retention are EPIC, FIPR, Statewatch
My position on the issue of traffic data retention and its impact on
civil society was presented at the first World Civil Society Forum, in
Geneva. The position paper
and the slides
are available.
G. Danezis, C. Diaz, S. Faust, E. Käsper, B. Preneel, and C. Troncoso. Efficient Negative Databases from Cryptographic Hash Functions, In Proceedings of the 10th Information Security Conference (ISC 2007), Lecture Notes in Computer Science LNCS, J. A. Garay (ed.), Springer-Verlag, 12 pages, 2007.
George Danezis, Stephen Lewis and Ross Anderson. How Much
is Location Privacy Worth?. Fourth Workshop on the Economics of
Information Security (WEIS
2005). Harvard University, 2 - 3 June 2005.
Steven J. Murdoch and George Danezis. Low-cost Traffic
Analysis of Tor. 2005 IEEE Symposium on Security and Privacy, May
8-11, 2005, Oakland, California, USA.
The social network analysis framework
used for "The Economics of Surveillance" research. Python code
including a map apply core, libraries for network ploting and KD-Trees,
and all the data.
Teaching
A few part II project proposals/MSc/Erasmus that I would like to
supervise can be
found at here. If
you are interested in a project related to anonymity feel free to
contact me.
I also gave a lecture
on anonymity for the Cambridge Part II computer security course.