Mathematical Methods for Homeland Spying

The amount of data relevant to home land defense is massive,distributed and growing rapidly through the addition of high volumedata streams and feeds. This presents fundamentally new mathematicalchallenges. These relate to: 1) the real time and near real timedetection of significant events in high volume data streams; 2) theforensic analysis of massive amounts of archived data to uncoverpatterns and events of interest; and 3) the mining of distributeddata, which for a variety of reasons will never be centrallywarehoused. To complicate matters further, homeland defense mustconcern itself with a variety of different data types, including,signals, text, images, transaction data, streaming media, web data,and computer to computer traffic.
http://cryptome.org/math-spy.htm