Boris Rozovsky studied painting before earning degrees
in mathematics. He leads a multidisciplinary team developing
a multi-target tracking system with potential military and
commercial uses, which will take advantage of the computer
board he holds here.
|
Hitting a Moving Target
Mathematician leads development of high-tech tracking system
By Carl Marziali
After 9/11, President Bush promised to “hunt down” Osama Bin Laden. But
the terrorist has proved to be an elusive target and White House
officials have admitted “we can’t be certain of where [Bin Laden]
is.”
A computer system capable of tracking hundreds or even thousands of
hostile agents — from terrorists to computer hackers — might help the
government locate someone like Bin Laden in the mountains of
Afghanistan. A USC College proposal to develop such a system recently
won a $5 million grant from the U.S. Department of Defense’s
Multidisciplinary University Research Initiative program.
Boris Rozovsky, professor of mathematics, leads the team of scientists
seeking to track large groups of moving targets automatically and in
real time.
“The development of multi-target tracking systems addresses a basic
national security need,” said Joseph Aoun, dean of USC College. “It is
notable that this outstanding project team is led by a theoretician.
This is a testament both to the relevance of pure mathematics to some
of our most pressing challenges, and to the degree to which USC has
succeeded in lowering barriers between fundamental and applied
research.”
Rozovsky, who directs the Center for Applied Mathematical Sciences
(CAMS) in the College, co-developed the general nonlinear filtering
theory and applications in the 1980s and 1990s. A filter is a
mathematical tool for detecting and tracking motion in a chaotic
environment. Linear filters fail when the motion becomes complex and
erratic. Rozovsky’s nonlinear filtering method overcame that problem.
Among other applications, nonlinear filters can be used to track
targets changing directions in high background noise, such as cruise
missiles in the sky or intruders moving through rugged terrain.
“It’s the problem of finding a needle in a haystack, but the problem is
the needle is moving,” said Rozovsky, who holds a joint appointment in
aerospace and mechanical engineering at the USC Viterbi School. “Or it
might not be a (single) needle, but many with different
characteristics.”
So far, the number of targets that can be followed at one time has been
limited to a few dozen. The current project aims to achieve real-time
tracking of thousands of mobile agents.
Rozovsky called this a “quantum leap” over previous systems, made
possible by advances in computing power and basic mathematical
research.
“Now you could think about tracking a large number of terrorists,” he said. “The speed is almost there.”
The same system could be used to detect security breaches in computer
systems, he said, calling this an urgent problem because each advance
in information technology tends to outstrip the corresponding advance
in security systems. Other applications could include tracking drugs,
blood clots or other substances as they move through the body.
The team’s winning proposal, titled “Spatial-Temporal Nonlinear
Filtering with Applications to Information Assurance and Counter
Terrorism,” beat six other finalists.
“My main achievement is that I was able to collect a brilliant team,”
Rozovsky said. “Every person is the best in the world in his or her
particular area.”
Team members are USC’s Paul Cohen of the Information Sciences Institute
and mathematician Alexander Tartakovsky — both members of CAMS — as
well as computer scientists Christos Papadopoulos and Isaac Cohen,
formerly of USC. Other team members are Andrea Bertozzi, Jeffrey
Brantingham and Tony Chan of UCLA, and Venugopal Veeravalli of the
University of Illinois at Urbana-Champaign.
|
|
|
