This paper deals with higher order approximation for discrete-time systems. It is shown that approximated feedback linearization at the second order can always be achieved under feedback compensation based on an approximated observer. An example is given in order to illustrate the control design and the efficiency of the proposed method.