Investigation of Data Fusion for Nuclear Threat Assessment, 20-R9723

Printer Friendly Version

Principal Investigators
Olufemi Osidele
Thomas Glass
David Turner
Robert Rogers
Deborah Waiting

Inclusive Dates:  07/01/07 – 07/01/08

Background - To address current domestic security concerns, many national studies have called for research on technologies for integrated information management. A recent National Research Council study found that large volumes of data from diverse sources must be acquired, integrated, and interpreted to support counterterrorism efforts. The study recommended advancing the practical utility of data fusion for intelligence analysis. Data fusion combines multisource, multi-format data to support timely interpretation and efficient use of information. Every year, about three million packages of radioactive materials are shipped across the country by road, rail, air, and waterways. A large portion of these packages are transported between nuclear facilities. In other sectors of business and commerce, millions of shipments of hazardous materials and critical goods, including food supplies, traverse the nation by various modes of transportation. From a security standpoint, these shipments are exposed to threats of theft, misdirection or malevolent use.

Approach - Simulation-based data fusion methods for detecting security threats to potentially vulnerable assets were investigated. The methods convert multisource intelligence data items into standardized data objects and combine these objects with static and dynamic operational data. Once the data are combined in this way, analysts can define and apply various criteria for detecting potential threats. Interfaces with geographical information systems were examined to support data acquisition, transport simulations, and threat visualization. Various intelligence and operational data sources were investigated and linked to scenarios depicting intercity shipment of nuclear materials. Simulation experiments were conducted to test the data fusion methods.

Accomplishments - Experiments with hundreds of shipments and thousands of intelligence data items achieved more than 95 percent threat detection accuracy. This successful proof of concept led to developing a prototype analysis platform (Hydra) for threat detection and assessment. Hydra incorporates five innovative data fusion methods, a supporting system architecture and data processing stream. These features are included in a pending United States patent application.

Section of Hydra Map Window showing standardized intelligence data (colored circles), static operational data (facility location, roads, railways, and waterways), and several hypothetical dynamic events.
Section of Hydra Map Window showing hypothetical threat alert (triangle) triggered by four intelligence data types at a population center. Inset is the alert popup window with details of the alert.

2008 Program Home