Blind Identification of Nonlinear Channels: A Subspace Approach

GÓMEZ, JUAN CARLOS; BAEYENS, ENRIQUE

Santa Fe

Simposio; Simposio Argentino de Tecnología (AST2008) (ISSN 1850-2784); 2008

SADIO: Sociedad Argentina de Informática e Investigación Operativa

A subspace based approach for the blind identification ofnonlinear Single Input Multiple Output (SIMO) time-invariant chan-nels is presented in this paper. The considered models include, but arenot limited to, polynomial approximations of nonlinear channels (i.e.,Volterra models), which have been used to represent satellite communications, magnetic recording channels, and physiological processes, amongother applications. The channel parameters are estimated in closed form,up to a scalar factor, by computing the column space of the output Hankel matrix, via a Singular Value Decomposition (SVD), and then solvinga Least Squares (LS) estimation problem also resorting to an SVD. Numerical simulations are used to illustrate the performance of the proposedalgorithms.