Neural network modeling and identification of nonlinear MIMO channels

Abstract

The paper proposes a neural network (NN) approach for modeling and identification of a class of nonlinear multiple-input multiple-output (MIMO) channels. The unknown MIMO system is composed of a set of single-input memoryless nonlinearities followed by a linear combiner. The proposed NN model consists of a set of single-input memoryless NN blocks followed by an adaptive linear combiner. The performance of the proposed scheme is shown to outperform the classical multi-layer perceptron (MLP) in terms of convergence speed, mean squared error (MSE) and computational complexity. For uncorrelated inputs, the proposed NN structure enables the identification of each of the unknown nonlinearities as well as the combining matrix. Several simulation results and applications are presented in the paper, including tracking of slowly time-varying MIMO channels, and fault detection and characterization in nonlinear MIMO systems.