An adaptive technique for modeling second-order Volterra systems with sparse kernels

Abstract

This paper proposes a simple technique for modeling second-order Volterra systems using an adaptive second-order Volterra delay filter (ASOVDF). The developed filter structure essentially extends an adaptive FIR delay filter to include linear and quadratic filter coefficients with an input assumed to be a zero-mean i.i.d. sequence with a symmetric distribution. The implementation of the ASOVDF is based on a stage-by-stage modeling process. At each stage, a dominant delay element is determined, the corresponding adaptive filter coefficient is incorporated to the adaptive filter coefficients from previous stages, and then these filter coefficients are adapted via the recursive least-squares algorithm. The ASOVDF requires few filter coefficients and has better performance and less computational complexity over the conventional adaptive second-order Volterra filter (ASOVF) in modeling second-order Volterra systems with sparse kernels.