A Null-Zone Decision Feedback Equalizer Incorporating Maximum Likelihood Bit Detection

A receiver structure is described, for use in pulseamplitude modulation (PAM) data communication systems, which combines null-zone detection and maximum likelihood (ML) processing to improve upon the performance of a conventional decision-feedback equalizer (DFE) while retaining a manageable level of complexity. If the output of the equalizer does not fall in the null zone, the receiver operates as a conventional DFE; however, when the equalizer output falls in the null zone, the decision on that data symbol is deferred one sampling instant and the two most likely signal levels are fed back into the equalizer, generating two competing outputs at the next sampling instant. Applying ML bit detection to this pair of outputs, and optimizing the width of the null zone, improves the effective signal-to-noise ratio of the receiver. Deferring the bit decision beyond one bit interval permits the utilization of additional received samples, and this extended receiver further increases the effective signal-to-noise ratio. The penalty paid for using a null zone, as opposed to performing ML detection on each bit, is at least a 3 dB loss of the usable (by a Viterbi algorithm (VA) receiver) energy contained in the interfering pulse samples.