Active sensing policies for slack-descriptor systems

Digital networks and the capability of addressable sensing devices offer new challenges for system theory. Rather than reading sensors continuously or periodically, as has traditionally been done, addressable sensing devices allow readings to be taken aperiodically and only as needed. In this paper, we introduce the notion of a sensing policy to determine which sensors should be read and when. The policy is dynamic, such that sensing times are determined from the current state estimates and the goals of the sensing policy. Sensing policies are considered in the context of slack-descriptor system models, which allows the representation of uncertain dynamics and sensor uncertainty through linear constraints.