Event-based joint estimation for unknown inputs and states: A distributed recursive filtering method

This article addresses the problem of distributed joint estimation for a class of discrete-time time-varying systems subject to random nonlinearities and unknown inputs over sensor networks. For the purpose of energy-saving, the dynamic event-triggering mechanism is adopted to govern the signal transmission between the sensor and the local estimator. First, some constraint conditions are introduced to decouple the unknown input to eliminate their impact. Then, by means of mathematical induction, an upper bound of the filtering error covariance is individually obtained for the state and the unknown input by solving coupled Riccati-like difference equations. Subsequently, the matrix simplification method is adopted to tackle the sparsity problem caused by sensor networks. In addition, the required distributed estimator gains are acquired by minimizing the obtained upper bounds of filtering error covariances. Finally, a numerical simulation is given to illustrate the effectiveness of the proposed joint estimator design scheme.

Automated summary

This article addresses the problem of distributed joint estimation for a class of discrete-time time-varying systems subject to random nonlinearities and unknown inputs over sensor networks. A dynamic event-triggering mechanism is adopted to govern signal transmission between the sensor and the local estimator for energy-saving purposes. Constraint conditions are introduced to decouple the unknown inputs and eliminate their impact. An upper bound of the filtering error covariance is obtained for the state and unknown input using mathematical induction and solving coupled Riccati-like difference equations. Required distributed estimator gains are acquired by minimizing the obtained upper bounds of filtering error covariances. A numerical simulation demonstrates the effectiveness of the proposed joint estimator design scheme.