Nonlinear Modeling of Super-resolution Near-Field Structure

Abstract

A correct and accurate model of optical storage systems is very important in the development and performance evaluation of various data detection algorithms. In this paper, we present a nonlinear modeling of a super-resolution near-field structure (super-RENS) read-out signal using the neural network and second-order Volterra models. The experiment results indicate that the nonlinear autoregressive network with an exogenous input (NARX) and the Volterra models considered in this study are superior to the normalized least-mean-squares finite impulse response (NLMS FIR) adaptive filter, the use of which is a linear modeling approach. We verify that the neural network and Volterra models can be utilized for the nonlinear modeling of super-RENS systems. Nonlinear equalizers may be developed on the basis of the information obtained from this nonlinear modeling. (C) 2009 The Japan Society of Applied Physics