CCSP Seminar: Sirin Nitinawarat, "Controlled Sensing for Sequential Multihypothesis Testing"
Wednesday, September 18, 2013
5:00 p.m. 2168 AV Williams Bldg.
For More Information:
Prakash Narayan prakash@umd.edu

Communications, Control and Signal Processing Seminar Controlled Sensing for Sequential Multihypothesis Testing

Sirin Nitinawarat Postdoctoral Research Associate Coordinated Science Laboratory University of Illinois, Urbana-Champaign

Abstract The problem of multiple hypothesis testing with observation control is considered in the sequential setting. In the case of uniform sensing cost, a sequential test is proposed generalizing the one studied in earlier work by Chernoff for binary hypothesis testing. Using the notion of decision making risk in place of the overall probability of error, the proposed test is shown to be first-order asymptotically optimal for multihypothesis testing {\em in a strong sense.} Another test is also proposed to meet {\em distinctly predefined} constraints on the various decision risks {\em non-asymptotically,} while retaining asymptotic optimality. Then a new model for controlled sensing for sequential multihypothesis testing is proposed.

This new model generalizes the existing one in two aspects. First, it includes a more complicated memory structure in the controlled observations. Second, it allows for a general cost structure which entails accumulating up to the final decision time the total control cost with respect to an arbitrary cost function. Consequently, this new model is relevant to a wider class of applications, particularly in distributed and mobile sensor networks. A sequential test is proposed for this new model and is shown to be strongly asymptotically optimal. Interestingly, it is shown that the optimal causal control policy for the controlled sensing problem is {\em self-tuning,} i.e., maximizing an inherent "inferential" reward simultaneously under every hypothesis, with the maximal value being the best possible, even when the true hypothesis is known at the outset.

Joint work with Dr. George Atia and Professor Venugopal V. Veeravalli.

Biography Sirin Nitinawarat obtained the BSEE degree from Chulalongkorn University, Bangkok, Thailand, with first class honors, and the MSSE degree from the University of Wisconsin. He received his Ph.D. degree from the Department of Electrical and Computer Engineering and the Institute for Systems Research at the University of Maryland, College Park, in December 2010. He is now a postdoctoral research associate at the Coordinated Science Laboratory at the University of Illinois at Urbana-Champaign. His research interests are in stochastic control, information and coding theory, statistical signal processing, estimation and detection, communications, and machine learning.

This Event is For: Graduate • Undergraduate • Prospective Students • Faculty • Post-Docs • Alumni