Dr. Yaakov Bar-Shalom, Marianne E. Klewin Professor in Engineering and a Board of Trustees Distinguished Professor of Electrical & Computer Engineering, has received a $630,000 three-year grant from the U.S. Department of Defense to develop practical multi-target tracking and multi-sensor data fusion algorithms that will aid the U.S. military in accurate detection and characterization of targets in the field. The work is expected to enhance the efficiency of surveillance systems when it is deployed for use with domestic and overseas radar systems.
The research will draw upon research underway in Dr. Bar-Shalom’s lab, which involves estimation and statistical decision theory combined with mathematical optimization. In collaboration with his colleagues, Electrical & Computer Engineering professors Peter Willett and Krishna Pattipati, Dr. Bar-Shalom will develop algorithms that accept input from multiple and diverse sensors; reduce clutter and noise – as well as individual sensor biases – to optimize reliability; send the sensor data to a central point where it can be merged and organized to reveal an accurate “single integrated” picture of multiple targets simultaneously; and trigger an appropriate response. The algorithms will be used in large computers housed either on ships or air command centers to track both aircraft and ballistic missile targets, and possibly land or sea targets.
The primary objective of the research will be to create a single integrated air picture or SIAP – a networking tool that will permit field personnel at different locations to “see” the same picture of the battlespace, including allied and enemy units on land, sea and in the air. To achieve this goal, Dr. Bar-Shalom said the algorithm must overcome several systemic problems associated with communication networks, including the procedure for handling out of sequence measurements (OOSM), which are sensor observations or measurements that arrive at the data fusion center not in the sequence in which they were obtained, but rather out of order due to a variety of delays. The team will seek to develop ways to incorporate the incoming data in the target track estimates without having to re-order them, which could be excessively time-consuming. In large systems, he said, the computational requirements can become prohibitive because of their exponential growth with the size of the problem.
Dr. Bar-Shalom commented that the team will publish their results in leading peer-reviewed journals to ensure the greatest integration of the algorithm into defense applications. “The algorithms I developed that are used in a large number – more than 50 – of Raytheon radars were picked up from my open literature papers and short courses I offered.”
A recognized international expert in target tracking, Dr. Bar-Shalom is credited with originating the probabilistic data association filter (PDAF) for target tracking in a low signal-to-noise ratio environment; pioneering the theoretical information limit for estimation in the presence of false measurements – and an algorithm that meets this limit; and developing the optimal track-to-track fusion (TtTF) equations for real-world asynchronous decentralized surveillance systems. These tools and tracking paradigms are used worldwide for target detection and tracking by military and national defense organizations.
