A robust FIR filter for linear discrete-time state-space signal models with uncertainties

This letter proposes a robust finite impulse response (FUR) filter for a class of linear discrete-time systems with quadratic bounded uncertainties. A set of all reachable current states under uncertainties is obtained from inputs and outputs measured on a recent finite time interval, which is represented in an ellipsoidal form. To minimize the maximum estimation error due to uncertainties, the center of the ellipsoid is chosen as an estimated state and the corresponding estimation error is obtained from the major axis of the ellipsoid. It is shown by simulation that the proposed robust FIR filter has a more robust performance than both robust infinite impulse response (IIR) filters and nominal FIR filters.