Blind Equalization for Chaotic Signals Based on Echo State Network and Kalman Filter Under Nonlinear Channels

Abstract

For nonlinear channels, we propose a new blind equalization method for chaotic signals based on echo state network(ESN) and Kalman filter(KF) in this letter. Using the short-term predictability of chaotic signals, ESN, a new feedback neural network, is used to approximate unknown chaotic maps and predict the output of the equalizer. Furthermore, KF combined with the finite impulse response(FIR) is used as the equalizer in this letter, and its parameters are adjusted by minimizing the prediction error of ESN. Simulation results show that compared with other blind equalization methods for chaotic signals, the proposed method has the advantages of smaller mean square error and faster convergence rate, and is suitable for strong nonlinear channels.