Tryst: The Case for Confidential Service Discovery Jeffrey Pang
(CMU), Ben Greenstein (Intel Research Seattle), Srinivasan Seshan
(CMU), and David Wetherall (Intel Research Seattle, University of
Washington)

Summary: Today Service Discovery is done in unsecure and non-confidential way;
it has uses because it makes it easy.  However it divulges information that you
may not want to reveal.  You will announce identity, and information that can
others can use for exploits, which leads to keeping of history which can be
correlated with location.  Finally SD today is unauthenticated.

Jeff also described the challenges in design of Tryst; in particular,
Key exchange, and establishment and trust.  And how Tyrst solves some
of these.

Q & A:

Q: The symmetric keys in your system are pair-wise, would that
require many exchanges. 

A: Overhead might be reasonable because many devices do not have so
many associations.

Q: Just looking at the side information can reveal information.

A: Yes there are plenty of other channels that can reveal
information, for example, timing and size of messages, or physical
proximity that can reveal information.