Nonlinear system compensation based on orthogonal polynomial inverses

Abstract

Nonlinear compensation using power series inverses suffers from various limitations, resulting from the coupling between the nonorthogonal basis elements used to represent the nonlinearity. These problems can be largely overcome by using representations of the nonlinearity and its inverse, based on orthogonal representations. We give a brief discussion of the disadvantages of the power series compensation technique. We then present a technique for composing and inverting orthogonal representations. Using this inversion technique, we derive a nonlinear compensation method that is based on the removal of specific terms in the orthogonal series. Compared with the power series method, this provides much better cancellation, and in some cases complete removal, of selected distortion terms. We illustrate the technique using examples involving narrow-band and broad-band inputs. We then investigate the robustness of the orthogonal compensation approach to variations in input signal properties.