Event-triggered state estimation in vector linear processes

Lichun Li, Michael Lemmon, Xiaofeng Wang

Abstract: This paper considers a distributed estimation problem in which a sensor sporadically transmits information to a remote-observer. An event-triggered approach is used to trigger the transmission of information from the sensor to the remote-observer. The event-trigger is chosen to minimize the mean square estimation error at the remote-observer subject to a constraint on how frequently the information can be transmitted. This problem was studied by O.C. Imer et al. [1] and M. Rabi et al. [2] where the observed process was a scalar linear system over a finite time interval. This paper extends those earlier results by relaxing the prior assumption that the initial condition is zero-mean with no measurement noise. It extends those earlier results to vector linear systems through a computationally efficient way of computing sub-optimal event-triggering thresholds.