Blind identification of Laguerre systems

Abstract

Laguerre filters have infinite impulse responses but with finite tapped delay-line parameterizations. This paper investigates subspace-based blind identification of single-input, multiple-output (SIMO) Laguerre filter tap coefficients, the internal filter state and the input, given only noisy observations of the output. A state space model for the SIMO Laguerre system is derived from which blind estimation algorithms are developed. The Laguerre filter tap coefficients can be estimated from the column space of a certain Hankel matrix constructed from noisy output observations while the internal state and input can be estimated from the row space by exploiting state space structure.