Adaptive identification and equalization of magnetic recording channels

Abstract

AbstractA new RLS adaptive Volterra filter is presented. The nonlinear filtering problem is transformed into an equivalent multichannel, but linear, filtering problem. The multichannel input signal is completely orthogonalized using sequential processing multichannel lattice stages. Thus, a fast convergent, highly modular and, simple filter with good numerical properties is designed. In the identification of magnetic recording channels, the filter identifies the channels directly and parameters for the channel nonlinearity are quantified simultaneously. In the equalization of magnetic channels, the most effective equalizer length can be assigned dynamically.