Convergence analysis of an algorithm for blind equalization

The authors consider a key algorithm for blind equalization and derive expressions for the evolution of the equalizer output error trajectory. The authors develop a model to examine the convergence behavior of this algorithm. Suitable approximations are incorporated into the model to facilitate analysis. The validity of these approximations is demonstrated for a typical communication channel. It is shown that even for a channel for which the assumption of Gaussianity of the equalizer input data may not be very good, the analysis presented predicts the convergence behavior reasonably well.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">></ETX>