Nonlinear equalizer for correcting intersymbol interference in a digital

Abstract

A non-linear equalizer corrects for intersymbol interference in a digital data transmission system by introducing baud rate samples into an N-stage tapped delay line and provision is made for storing a sequence of M prior decisions in a prior decision register. Both the samples from the tapped delay line and the contents of the prior decision register are coupled to a weighting matrix processor which is updated to permit adaptation to different channels conditions. When initially placed into operation, a training sequence is coupled to a reference register and differences between arrived-at decisions and the true symbols are employed to provide an error vector which is then used to update the weighting matrix in the processor. Decisions on the data estimates are hard-limited to form a set of M prior decisions which are placed into the prior decision register. The training sequence is employed over a suitable acquisition time so that the weights of the processor matrix will have stabilized at values suitable for the channel through which unknown digital data will be transmitted. Because the equalizer of the invention is non-linear, it can remove intersymbol interference without the severe noise enhancement problem of linear equalizers on some channels. Moreover, not only can it remove intersymbol interference from both precursors and postcursors, but it has the ability to modify previously made decisions upon the receipt of additional data from the channel.