Comments on "Channel Identification for High Speed Digital Communications"

Abstruct-It is shown that the accelerated equalizer convergence reported in the above paper,’ obtained by using a second-order adaptive algorithm, is misleading. The second-order algorithm presented is approximately equivalent to the usual fit-order algorithm with a larger adaptation constant (step-size). In the above paper,’ Proakis derives a second-order (and also a third-order) adaptive algorithm based on linearly filtering the gradient with a single-pole low-pass filter? The algorithm then becomes