Two fusion predictors for multisensor discrete time linear system

Abstract

New multi-step fusion predictors for discrete-time linear dynamic systems with different types of observations are proposed. The multi-step fusion predictors are formed by a summation of the local Kaiman filters/predictors with matrix weights depending only on time instants. According to fusion sequence, there could be two kinds of multi-step fusion predictors; the relationship between these fusion predictors is established. Then, the accuracy and computational efficiency of the fusion predictors are demonstrated on a first-order Markov process and a ground moving target indicator model with multisensor environment.