Distributed Unknown Input Observer

In this article, a distributed observer is designed for the state reconstruction of continuous-time linear dynamical systems with unknown inputs. Specifically, a group of sensor nodes are employed to measure the output of such a system. Due to its limited measurement, a single node may fail to reconstruct the full state. By acting on a consensus cooperative strategy, however, each node can produce an accurate state estimate, regardless of the specific input signal. The attractive performance of distributed unknown input observer relies on nothing but common neighboring communication, as well as an adequate local measurement that overlaps the effect of the unknown input along system dynamics. Correspondingly, the required network connectivity and rank conditions are given, with their linkages to the classic result being discussed.