Conjugate Unscented Transform based approach for dynamic sensor tasking and Space Situational Awareness

This paper presents a framework to manage sensor systems for the effective characterization and monitoring of uncertainty in distributed sensing problems. The problem of Space Situational Awareness, where a multitude of Resident space objects has to be tracked by a limited number of sensors is taken as the primary example. The central motive of the paper is to iteratively update sensor modalities of the sensors in an efficient manner so as to improve the overall tracking process of the dynamic targets. Recently developed Conjugate Unscented Transformation (CUT) method has been utilized to efficiently compute the expectation integrals that arise in the nonlinear state estimation process and sensor performance evaluations. Few sensor optimization problems are posed to optimally pair sensors to the targets using information measures such as Fisher Information and Mutual Information. Finally, numerical simulations illustrate the effectiveness of the proposed methodology.