A nonlinear filtered-x prediction error method algorithm for digital predistortion in digital subscriber line systems

Digital predistortion of nonlinear systems is an important topic in many practical applications. This paper considers direct predistortion of a Volterra system by connecting in tandem an adaptive Volterra predistorter. The coefficients of the predistorter can be recursively estimated using the nonlinear filtered-x least mean squares (NFxLMS) algorithm. In this paper, the prediction error method (PEM) is used to derive a novel nonlinear filtered-x PEM (NFxPEM) algorithm. A simulation study on Volterra systems shows that the NFxPEM algorithm more significantly suppresses spectral regrowth and converges much faster than the NFxLMS algorithm. Also, the NFxPEM algorithm is used in this paper to design more efficient digital predistorter—as compared with the NFxLMS algorithm—for digital subscriber line systems. Copyright © 2011 John Wiley & Sons, Ltd.