Skip to Main Content
This paper is devoted to the design and performance analysis of an autonomous decentralized monitoring system that degrades gracefully under natural or malicious sensor malfunctioning. In the scenario considered, the sensors are characterized by a quantity referred to as data quality, and a method for estimating process variables, based on sensor measurements and data quality, is developed. Using these estimates, the plant status assessment is carried out, and the entropy of the resulting probability mass function, augmented by symmetrized relative entropy of sensor measurements, is used to drive the so-called rational controllers, which force the monitoring system to operate in the optimal state. Along with analytical results, the paper presents numerical examples, which illustrate the efficacy of the resulting resilient monitoring system, using a metric based on the Kullbeck-Liebler divergence.