Least-squares channel equalization performance versus equalization delay

Linear channel equalization has been a successful way to combat intersymbol interference (ISI) introduced by physical communication channels at high e nough symbol rates. We consider the performance of least-squares equalizers in the single-input/multi-output (SIMO) channel context when the true chan nel is composed of an mth-order significant part and tails of "small" leadi ng and/or trailing terms. Using a perturbation analysis approach, He show; that if the diversity of the significant part is sufficiently large with re spect to the size of the tails, then the lth-order least-squares equalizers , with l greater than or equal to m - 1, perform well for all the delays co rresponding to the significant part. On the other hand, the performance of the equalizers for the delays corresponding to the tails may be poor.