Dynamic threshold implementation of the maximum-likelihood detector for the EPR4 channel

Abstract

The authors show that the maximum-likelihood Viterbi detector for the binary extended class-4 partial response (EPR4) channel, in additive Gaussian noise, can be implemented using dynamic thresholds. The dynamic threshold method uses eight thresholds, whereas the add-compare-select method uses eight state metrics. The dynamic threshold method has the same performance as the add-compare-select method, and is presented as an alternative to that method. For selected implementations of the two methods, the dynamic threshold method was found to use 20% less adders than the add-compare-select method. In addition, the path memory for the EPR4 maximum-likelihood Viterbi detector, using either implementation, can be reduced by up to 37.5%. >