An asynchronous technique for distributed estimation based on zonotopes

This paper deals with the problem of guaranteed distributed estimation under an asynchronous communication scheme. The objective is to estimate the state of a perturbed linear plant using a set of agents which communicate using a common network. The agents have access to a limited set of system outputs affected by bounded noises, so collaboration among them is needed. Both measurement noises and disturbances are bounded, but no stochastic distribution is assumed to be known. This work proposes a guaranteed estimator based on the use of zonotopes, which constitutes a useful way to represent the estimated set. The transmissions between agents are event-triggered. A novel triggering rule is presented. It weights, on the one hand, the error with past transmissions, and on the other, the uncertainty of the estimation. Simulation examples illustrate the achieved results.