Expert in the Spotlight: Stefano Coraluppi

In this series, we introduce a scientist, who makes use of signal processing techniques for his research findings, by means of an interview. This month, we are happy to introduce Dr. Stefano Coraluppi from Systems & Technology Research, whose research interests mainly focus on multi-target tracking.

What is your application that may require signal processing technologies?

In graduate school, I was fascinated by estimation and control theory, and my PhD thesis was in stochastic control. Early in my professional career, I was involved in a project on multi-target tracking (MTT) for ground surveillance with a variety of sensors. Since that time, I have worked in MTT with multiple sensors in numerous domains including ground, undersea, maritime, and air surveillance. I have also applied these techniques to cell tracking problems. I like to view MTT and related sensor management tasks as generalizations of the estimation and control technologies that I studied of my youth. Interestingly, many of the well-known researchers in MTT come from a background in control theory.

Generally, MTT algorithms operate with detection-level data, thus relying on signal processing technologies that transform raw sensor data into contacts (sonar terminology) or plots (radar terminology). Most MTT algorithms operate in the time domain. Perhaps surprisingly, some well-known results in estimation theory, e.g. the optimality of centralized and sequential processing, do not generalize directly to the multi-object setting.

In your opinion, are there any emerging applications that signal processing technology may play an important role?

Signal processing advances enabling MTT algorithms to operate in the encrypted domain would allow for collaborative surveillance among honest-but-curious sensing platforms. This would have direct applicability to many current challenges in defense and security.

In your opinion, what are the most exciting NEW techniques in signal processing field?

Stochastic methods for nonlinear filtering, including particle filtering, have been around for some time. These prove to be useful for highly nonlinear non-Gaussian estimation problems that are often encountered in MTT. A more recent and very exciting development has been the introduction of stochastic sampling methods based on Markov Chain Monte Carlo (MCMC) methodologies to address the data association problem, the key challenge in MTT. A seminal paper in this area is due to S. Oh, IEEE Transactions on Automatic Control, March 2009. This opens up many possibilities to apply stochastic sampling methods to MTT. A direct comparison of such methods to track-oriented multiple-hypothesis tracking (TO-MHT), the leading methodology for MTT, remains to be investigated.

In which way have you been connected first with IEEE SPS (university, conference, etc…)?

Years ago I was a member of IEEE SPS, but for many years now I have only been affiliated with the IEEE Aerospace and Electronic Systems Society (AESS), in which I am quite active – and time is limited, unfortunately! I attended and gave a plenary at the IEEE SAM 2012 conference, which was quite enjoyable. Many of my research colleagues are active in SPS, so I don’t feel too far removed from it. Often, key papers in MTT are published in IEEE TSP.

Are you member of other IEEE Societies? If yes, could you compare through a selected application the two societies, e.g. how to approach a given problem, are they more application or methodology oriented, practical/theoretical etc.

AESS tends to have a fairly applied focus. Nonetheless, there are key algorithmic challenges which overlap with research areas in signal processing, information theory, and automatic control.

Which application fields should be more focused in IEEE SPS publications?

SPS publications have historically been very strong, so I don’t have any specific advice to impart. I would encourage those with problems that involve data association and measurement provenance uncertainty to be aware of the growing MTT literature. In the old days, MTT techniques were less mature and at times somewhat ad hoc when compared with more established fields like signal processing and control. This is far less true today, and there is a wealth of exciting new algorithmic developments and performance metrics that may find application in more traditional signal processing challenges.

Do you have any suggestions about our e-NewsLetter? 

I have only recently become aware of the e-NewsLetter. It is informative and quite readable, so I plan to keep an eye on it going forward. I would encourage SPS members to do the same with publications of sister societies in the IEEE, as there is significant overlap in research challenges and ideas. Much can be learned through cross-fertilization.

Brief Bio

Stefano Coraluppi received the BS degree in Electrical Engineering and Mathematics from Carnegie Mellon University in 1990, and MS and PhD degrees in Electrical Engineering from the University of Maryland in 1992 and 1997.  He has held research positions at ALPHATECH Inc. (1997-2002), the NATO Undersea Research Centre (2002-2010), Compunetix Inc. (2010-2014), and Systems & Technology Research (since 2014).  His research interests include multi-target tracking and multi-sensor data fusion for defense and security. He is Secretary of the International Society of Information Fusion (ISIF), Associate Editor-in-Chief for the ISIF Journal of Advances in Information Fusion, Technical Editor for Target Tracking and Multisensor Systems for the IEEE Transactions on Aerospace and Electronic Systems, and a Senior Member of the IEEE. He served as General Co-chair of ISIF/IEEE FUSION 2006 and Technical Program Co-chair of ISIF/IEEE FUSION 2014. This year, he serves as Technical Co-chair of ISIF/IEEE FUSION 2015. He lectures regularly for the NATO Science and Technology Organization.