Its framework allows one to establish the connections between the feedback linearization method and one-step-ahead prediction horizon, but its implementation does not require on-line optimization algorithms.