Knowledge Discovery from Sensor Data for Security Applications

Evolving threat situations in a post-9/11 world demand faster and more reliable decisions to thwart the adversary. One critical path to enhanced threat recognition is through online knowledge discovery based on dynamic, heterogeneous data available from strategically placed wide-area sensor networks. The knowledge discovery process needs to coordinate adaptive predictive analysis with real-time analysis and decision support systems. The ability to detect precursors and signatures of rare events and change from massive and disparate data in real time may require a paradigm shift in the science of knowledge discovery. This chapter describes a case study in the area of transportation security to describe both the key challenges, as well as the possible solutions, in this high-priority area. A suite of knowledge discovery tools developed for the purpose is described along with a discussion on future requirements.