On the equalization of nonlinear systems

The Volterra series expansion is widely used for the identification and modeling of nonlinear systems. But few works can be found dealing with the equalization of nonlinear distortion. Based on the theoretical work of Schetzen (1980) this paper presents new practical equalization schemes. For discrete-time equalization an important question is how to choose the sampling frequency because nonlinear systems produce new, higher frequency components at their outputs. It is shown that for post-processing it is sufficient to sample at twice the maximum input frequency of the nonlinear system. The results are verified by simulations.