Dynamic Data-Driven Sensor Tasking with Applications in Space and Aerospace Systems

This chapter discusses the Dynamic Data-Driven Applications Systems (DDDAS)-based mathematical approaches associated with the problem of manag-ing a network of static and dynamic sensors to accurately characterize complex dynamical environments. Examples of sensor-network management include assign-ing a set of sensors for surveillance and tracking multiple ground/aerial/marine targets. Further examples include adaptive sensing of large-scale spatial phenomena such as weather, volcanic eruptions, and pollutants in air or large water bodies. By using information theoretic measures of sensor performance and the value of information predicted by the models, the formulation of optimally tasking a group of sensors to maximize mutual information is detailed. The resulting sensor-tasking optimization problem is shown to be combinatorial in nature, and its computational complexity expands with an increasing number of targets and sensors. Appropriate suboptimal approximations are presented to alleviate this computational complexity of the sensor-tasking problem. The submodular property ofthemutualinformationmeasureisutilizedtoprovideguaranteesontheoptimality of different approximations. Numerical simulations involve tracking of aerial and space objects with ground-based sensors and marine objects with sensors mounted on unmanned aerial vehicles. These simulations showcase the efficacy of optimizing the configurations of a limited number of sensor agents to better track multiple noncooperative targets.