Producing soft-decision information at the output of a class IV partial response Viterbi detector
K.J. Knudson, Jack K. Wolf, Larry B. Milstein · 2002
A method is described for producing soft-decision information at the output of a Class-IV partial response maximum-likelihood (PRML) detector, when the additive noise samples are independent, Gaussian random variables. This method directly applies the soft-decision Viterbi algorithm, presented by J. Hagenauer and P. Hoeher (1989), to the Class-IV partial response channel. The PRML detector is augmented to produce a reliability value for each binary symbol in the maximum-likelihood (ML) sequence estimate of the binary input. The reliability value is a function of the estimated probability of error for the binary symbol. The soft-decisions are used to enhance the performance of an outer error correction code.>