Decision-directed fractionally spaced equalizer control using time-domain interpolation

Abstract

The authors describe a time-domain Nyquist-rate algorithm that broadens the scope of fractional equalizers to include either least-mean-square control or polarity-dependent adaptation analogous to zero-forcing in synchronous equalizers. The study includes the analytic basis for interpolative control, a functional circuit description (with extension to symbol-error-polarity control), and a computer simulation that illustrates operation on a digital subscriber loop. The time-domain interpolative algorithm permits unique user-defined specification of the end-to-end Nyquist channel. Though the end-to-end channel may not be optimal, a priori specification affords a new dimension for creativity, possibly embracing novel baseband carrier and timing recovery schemes. The spectral uniqueness attained by satisfying the Nyquist-sampling condition seemingly reduces or eliminates the coefficient drift phenomenon. >