Fusion Performance Evaluation Based on Multisource-Multitarget Information Theory

Period of Performance: 05/22/2000 - 05/22/2002


Phase 2 SBIR

Recipient Firm

Scientific Systems Company, Inc.
500 West Cummings Park Array
Woburn, MA 01801
Principal Investigator

Research Topics


The ability to measure an algorithm's competence is a vital part in developing practical systems at all levels of data fusion. This aspect of metrology in data fusion is currently poorly understood and often relies on ad hoc techniques. e believe that a systematic and scientifically defensible approach to data fusion metrology at all levels 1, 2, 3, and 4 is now feasible, based on a direct generalization of information theory to the multisource-multitarget realm based on finite-set statistics (FISST). Phase I showed the feasibility assessment for fusion performance estimation for Level 4 fusion, and studied the effectiveness of different metrics. The Phase II objectives are to further develop, analyze, and refine the metrics by extending the results of Phase I to all levels of data fusion. Specific Phase II tasks are: (1) Refinement of Level 4 metrics, (2) Extension of the approach to Level 2 and Level 3, (3) Development of Measures of Robustness, (4) Adaptive Data Fusion, (5) Feasibility study of applying Genetic Algorithms to the fusion metrology problem, and (6) Development of Performance Evaluation Toolboxes. The project team includes Dr. Ronald Mahler of Lockheed Martin. Lockheed Martin will provide both technical and commercialization support in the application of the sensor fusion technologies.