Direct Linearization of Weakly Nonlinear Volterra Systems Using Adaptive Linear and Nonlinear FIR Filters

Abstract

Linearization of nonlinear systems is a very important topic in many practical applications. The linearization scheme of Gao and Snelgrove (1990) for weakly nonlinear Volterra systems using adaptive linear and nonlinear FIR filters is considered in this paper. The coefficients of these filters can be indirectly estimated using the Least Mean Squares (LMS) and Recursive Prediction Error Method (RPEM) algorithms as done in Gan and Abd-Elrady (2008b). In this paper, these coefficients are directly estimated using the Nonlinear Filtered-x Least Mean Squares (NFxLMS) algorithm and the Spectral Magnitude Matching (SMM) method. Simulation study shows that the suggested direct approaches can significantly suppress spectral regrowth and reduce nonlinear distortion as well as the indirect approach.