Nonlinear Equalization of the Super-Resolution Near-Field Structure Read-Out Signal Using the Adaptive Amplitude Nonlinear Gradient Descent Algorithm with the Sigmoid Function

Abstract

In this study, to mitigate nonlinearity in super-resolution near-field structure (super-RENS) read-out signals, we investigate nonlinear equalization based on the adaptive amplitude nonlinear gradient descent (AANGD) algorithm, which is suitable for processing nonlinear and nonstationary input signals with a large dynamic range. Note that the sigmoid function is employed as the nonlinear activation function since it is commensurate with the experimental data and entails simple implementation. The experiment results regarding bit error rate (BER), convergence speed, lookup table (LUT) size, and computational complexity show that the proposed equalizer outperforms the Volterra filter and linear finite impulse response filter with the normalized least-mean square (NLMS) algorithm.