Fast adaptive decision-selection equalizer convergence using a tree-structured algorithm

Previous research on adaptive equalization indicates that for some communications channels with nonlinear intersymbol interference (ISI) the decision state conventionally used as feedback can instead be used to select a detector model in order to achieve improved performance while maintaining simplicity compared to historic nonlinear solutions. For example, one method studied enhances the familiar adaptive decision feedback equalizer (ADFE) for binary transmission by using the past decision state to choose and adapt different sets of coefficients, i.e., different hyperplane detector boundaries. One notable drawback to the least-mean-squared (LMS) based adaptive decision-selection equalizer (ADSE) as implemented previously is decreased convergence rate. A tree-structured algorithm is proposed herein to speed convergence when the various conditional hyperplanes are not orthogonal in exchange for a manageable increase in memory and processing.