Fusion Predictors forMultisensor Discrete-Time Linear Systems

Two novelfusion predictors forlinear dynamic systems withdifferent typesofobservations areproposed. They areformedbysummingofthelocal Kalmanfilters/predictors withmatrixweights depending onlyon timeinstants. The relationships between themandtheoptimal Kalmanpredictor are discussed. Highaccuracy andcomputational efficiency ofthe fusion predictors aredemonstrated onthefirst-order Markov process andtheGMTIwithmultisensor environment.