Statistical Signal Processing

Website
Associated faculty: Castañón, Ishwar, Karl, Saligrama

The area of statistical signal processing focuses on the robust extraction of information in the face of uncertain data and models. Typically one is given a probabilistic description for an input signal and a model for a system that changes or distorts the signal. The resulting noisy observation is combined with these models to design algorithms for optimal processing. Fundamental issues include the nature of the basic probabilistic signal description, the model of the distorting system, the probabilistic description of the output signal given that of the input signal and the system, and the design of processing paradigms that are both appropriate and yet lead to tractable algorithms. Applications areas include communications, control, estimation, and decision making. Current research interests involve solution of inverse problems, decision making, recognition processing, and sensor networks.