Speaker: Hong-Sheng Zhou
Day:  Wednesday, 11/30/2005
Room: ITEB 201
Time: 3:30pm
Title: Private Searching on Streaming Data

Abstract:
The technique of private searching on streaming data is very important. It
allows the intelligence community to collect small fraction of potentially
useful information from very huge streaming sources of data under a
_secret_ sieving criteria. Note that we must keep the sieving criteria
classified; otherwise the adversaries could easily avoid their messages
from being collected by simply avoiding the criteria. 

Traditionally, we take the approach to collect all streaming data into a
secure environment, and then filter them under the criteria. Obviously no
information of the criteria will be leaked. However, it is very expensive
due to the heavy communication and the huge temporary storage.  A new
approach is to filter the data-streams at their sources. But, it is not
easy to keep the sieving criteria secret because the filtering codes could
fall into enemy's hands. 

In this paper, the authors solve this problem by using public key program
obfuscation over the filtering codes based on homomorphism encryption. I
will explain the new obfuscation technique and present the main scheme
based on Paillier encryption. 

Reference: 
Rafail Ostrovsky, William E. Skeith III: Private Searching on Streaming
Data. CRYPTO 2005: 223-240
http://dx.doi.org/10.1007/11535218_14