Adaptive filter theory (2nd ed.)
Adaptive filter theory (2nd ed.)
Hi-index | 0.00 |
LMS adaptive implementations of second order Volterra filters are subject to nonuniform convergence behavior analogous to the eigenvalue disparity effects in the conventional LMS. In the nonlinear filter however, this problem arises both due to the coloration of the data, and due to the impact of nonlinear operations on the data. In this paper we describe a novel lattice structure for a second order adaptive Volterra filter. This algorithm includes a two-stage adaptive orthogonalization which greatly reduces nonuniform convergence effects. The algorithm is relatively simple to implement and has a reasonable computational overhead.