Search and discovery in an uncertain networked world

This paper identifies two classes of sensor network problems that arise naturally in many applications. Fusion problem involves correlated data over the sensor network, while search/sensor selection involves selecting sensors that have useful information. The paper also describes an adaptive decision making system, which feeds back the previous decisions of the sensor network to make more informed decisions in the succeeding steps of the algorithm. The ideas that are developed for fusion problems proved to be useful in successfully executing the described algorithm. Problems such as target tracking in clutter and sensor selection can be fitted into the framework of this class.