Signal-dependent correlation-sensitive branch metrics for Viterbi-like sequence detectors

By applying the Euclidian branch metric in a Viterbi-like detector when the noise is signal-dependent and correlated, the receiver falls short of the maximum likelihood sequence detector (MLSD). We introduce new signal-dependent correlation-sensitive branch metrics for Viterbi-like implementations of the MLSD. We also provide an analytic analysis that calculates the probability of error of such a detector and compares it to the performance of the Euclidian detector. The new metric is well suited for magnetic recording applications where, especially at high recording densities, the noise is both correlated and signal-dependent.