Summary:

This paper presents a probabilistic key-sharing algorithm for distributed 
sensor networks in which a small number of randomly chosen keys are stored 
on each node.  By using more keys than schemes that use only a single 
"mission key," this key-distribution scheme allows for the compromise of 
some communication links without this causing the compromise of the entire 
network.  On the other hand, this flexibility is incorporated without 
requiring that each host share a unique key with every other host, which 
makes re-keying and the addition of new nodes to the network much easier 
than in schemes where each node shares a unique key with every other host.

The authors claim that this system allows for networks to remain connected 
with high probability even when nodes are compromised or keys expire and 
are revoked.  Results in graph theory and simulations run on random graphs 
are used to support these claims.


Pros:
  - Solution presented to a meaningful problem
  - Vast reduction in the amount of keys required to allow nodes to 
communicate with arbitrary nodes
  - Solution presented scales well
  - This solution limits the effect of the compromise of nodes

Cons:
  - Paper was missing depth -- lots of things touched on to quickly or 
glossed over
  - Key discovery is possibly energy inefficient
  - Lack of analysis of power costs and number of messages necessary to 
complete key discovery on different sized networks
  - Paper seems to rely on ability to discover compromise of nodes, but 
never presents a way to do this

Votes:
  - Strong Accept: 0
  - Accept: 13
  - Reject: 1
  - Strong Reject: 0


--
Adam Lee <adamlee@uiuc.edu>