A difference metric implementation for maximum-likelihood sequence estimation of binary input intersymbol interference channels

Maximum-likelihood sequence estimation (MLSE) is typically performed using the Viterbi algorithm to update state metrics. In this paper, we describe an equivalent method of performing MLSE detection by updating difference metrics rather than state metrics. The difference metric implementation is a generalization of the technique described by Ferguson (1972) for the (1-D) channel. The difference metric implementation applies to any binary input intersymbol interference (ISI) channel corrupted by independent and identically distributed Gaussian noise. For ISI channels with arbitrary coefficients, the difference metric implementation is significantly less complex than the state metric implementation. However, for partial response channels with integer coefficients, the reduction in complexity is not as dramatic.