Tensor-based methods for Wiener and Hammerstein channels identification

In this paper, we propose tensor-based methods for identifying nonlinear communication channels of Wiener and Hammerstein. For a Wiener channel, the parameters of linear subchannel are estimated using two different approaches based on the PARAFAC decomposition of the associated fifth-order Volterra kernel. The first approach is to apply the iterative ALS algorithm, while the second approach uses the SVD of the fifth-order Volterra kernel. For Hammerstein channel, we propose an approach based, also, on the fifth-order Volterra kernel. Then, the coefficients of nonlinear subchannels modeled as a polynomial, of both channels, are estimated by means of the RLS algorithm. The proposed identification methods is illustrated by means of simulation results.