Adaptive baseband digital communication and applicable convergence results

Several baseband adaptive communication systems are presented and analyzed with regard to 'self-synchronization' properties, intersymbol interference, and convergence properties. Recent convergence results, the proofs of which are contained in a companion report, are applied to provide extremely mild 'covariance decay-rate conditioners' for which the algorithms treated converge with probability 1. Of special interest are the convergence results treating correlated cyclostationary training data. Recent results on maximum likelihood sequence estimation are extended to treat the detection of general 'nonlinearly modulated' digital data over linear dispersive channels and nonwhite additive noise. Adaptive techniques for training the new detector structure are proposed for use when the channel and/or the noise covariance function are unknown.