We are addressing the "sender anonymity" problem,
which stated informally is the following: how does an individual
send a message anonymously and reliably to an adversary that controls
the communications network as well as some of the communicating
parties? More formally: how do two nodes conduct communications
such that no adversary -- either as a participant or outside observer
-- can prove to an arbitrator (that trusts that adversary's ability
to gather data) that a specific node, not in collusion with the
adversary, was the sender of a specific message? In essence, it
is the ability to distinguish the sender from the other, "innocent"
participants in the protocol.
There is currently a good deal of research into this problem,
with some interesting and somewhat useful results. [1], [2] However,
both the research dealing with providing the highest degree of
anonymity and the research dealing with efficient protocols that
provide some anonymity are flawed in that either they are good
(provide sufficiently high degrees of anonymity), are efficient,
or are practical, but never all of the above. Sometimes they are
none. We believe the problem has immediate relevance in today's
world of increasing censorship, especially on the Internet and
in technology in general. We are investigating solutions and in
the hopes of advancing the current state of knowledge.
[1] David Chaum, The Dining Cryptographers
Problem: Unconditional Sender and Recipient Untraceability,
Journal of Cryptology 1(1): 65-75 (1988).
[2] David Chaum, Untraceable electronic mail, return addresses,
and digital pseudonyms, Communications of the ACM, 24(2):
84-88, 1981.
Preliminary
Presentation (ppt),
(pdf)
Other Mini-PROBEs for Summer 2003
Algorithms for Facility Location
Designing Overlay Multicast Networks for Streaming
Dynamic Algorithms
HumanAUT
Moving Mesh Simulations
Space-Efficient
Point Location