The coefficients of the optimal steady-state -step ahead predictor for a scalar ARMAX process in general depend on . It is shown that a simple formula completely characterizes all these coefficients. This extends previous results on the characterization of ARMA predictors.
Keywords :
Autoregressive moving-average processes; Autoregressive processes; Equations; Frequency estimation; Instruments; Parameter estimation; Recursive estimation; Signal processing algorithms; Speech processing; State estimation; Steady-state;