Blind Equalizers By Techniques Of Optimal Non-Linear Filtering Theory

Blind deconvolution fundamental in signal processingapplications is a challenging problem. Althoughblind equalization is essentially a nonlinearfiltering problem it has not yet been treated by allthe well known advanced techniques of optimalnon-linear filtering theory. In this book we use theEdgeworth expansion maximum entropy argumentationsand the Laplace integral method in order to obtaintwo new groups of blind deconvolution methods whichoutperform the old and new algorithms. The proposedmethods are based on new closed formed approximatedexpressions for the conditional expectation suitablefor the blind deconvolution problem. Since theseexpressions do not impose any restrictions (exceptthat of even symmetric) on the probabilitydistribution of the (unobserved) input sequence theyare suitable for a wider range of source pdf comparedto Bellini's Fiori's or Haykin's expression. Thederivation of the above mentioned equalizers areaccompanied by theoretical analysis of theperformance in the mean square error (MSE) sense andare justified via simulation. These new methods areuseful to professionals in Communications who seekfor improved equalization methods.