Implementation-Friendly Constraint for Adaptive Finite Impulse Response Filters for Equalization

Abstract

In read channels in magnetic recording systems, a finite impulse response (FIR) filter is often used as an equalizer, and the filter coefficients are adapted via a least mean squares (LMS) algorithm. This equalizer adaptation loop can interfere with the timing recovery (TR) loop and the automatic gain control (AGC) loop that are also present in such channels. In a previous paper by L. Du et al., a linear constraint is proposed for the LMS loop so that it is decoupled from the TR and the AGC loops. In this paper, we generalize that constraint, and propose a new constraint that offers the advantage of easier implementation.