Channel memory truncation for maximum likelihood sequence estimation

Calculating the coefficients of a linear equalizer (prefilter) so that the combined response of the equalizer and a given channel is forced to a limited length (memory), while the signal-to-noise ratio at the equalizer output is maximized, is addressed. The process involves simultaneous calculation of a desired response that the combination of the channel and equalizer should approximate and the equalizer coefficients. An iterative algorithm which is an extension of our previous work (see Glbobecom'92, p.26, 1992) is first developed. This is followed by derivation of a non-iterative lattice-based algorithm which uses the Levinson algorithm, in part, to achieve a very low computational complexity.