About the Event
Abstract: Traditional inference takes measurements of a signal and then forms estimates of the region/signal of interest. In adaptive sensing, previous measurements are used to inform better allocation of resources in order to maximize performance, such as detection probability or estimation accuracy. In this work, sparse scenarios are considered where the interesting element is embedded in a much larger signal space. A framework for adaptive search for sparse targets is proposed to simultaneously detect and track moving targets with limited resources. Previous work is extended to include a dynamic target model that incorporates target transitions, birth/death probabilities, and varying target amplitudes. Policies are proposed that are empirically shown to have excellent asymptotic performance in estimation error, detection probability, and robustness to model mismatch. Adaptive sensor management is also studied in the context of developing fundamental performance limits for stable tracking of targets, such as the maximum number of targets that can be tracked, the maximum spatial uncertainty of those targets, and the system occupancy rates. Lastly, these tools are applied to a specific application, namely tracking targets using synthetic aperture radar (SAR) imagery. Due to the complexities of SAR imagery, a hierarchical Bayesian model is proposed for efficient estimation of the posterior distribution for the target and clutter states given observed SAR imagery. This model provides a unifying framework that combines working knowledge of the physical, kinematic, and statistical properties of SAR imagery.