Stability of receding horizon Kalman filter in state estimation of linear time-varying systems

The paper presents a state predictor for linear time-varying systems using Kalman filter with the receding horizon strategy. It can be seen as a standard Kalman filter which takes into account the most recent data, those included in a moving data window of fixed length. The main purpose here is to assure stability for this type of filter. Under standard conditions we can establish a minimum horizon length for which the closed-loop filter with the receding horizon gain is exponentially stable. The approach makes no direct reference to the properties of the underlying Riccati equation, which allow us to address more general problems that can not be coined in terms of Riccati equations.