Blind Equalization Based on Neural Network under LS Criterion by Gradient Iteration Algorithm

Abstract

A blind equalization based on neural network under LS criterion was proposed in this paper and gradient iteration algorithm adopted to avoid computing the reverse matrix of correlation of input signal. The BP algorithm in the traditional blind equalization based on feedforward neural network is a stochastic gradient descent algorithm, which has low convergence rate and high residual error; meanwhile, it is often absorbed in locally minimum. The method proposed in this paper has better performance and no adding computation complexity compare with BP algorithm. Simulation results show that the equalization performance is improved under the nonlinear communication channel condition.