Pattern-Dependent Noise Predictive Soft Detection in the Post-Processor With Error Filters

Abstract

This paper investigates a pattern-dependent noise predictive soft detection method for channel architectures that are based on a long target response and post-processing rather than a short target response and base-line wander compensation. We utilize properties of the autoregressive pattern-dependent noise model to compare tentative Viterbi sequence with alternative sequences in the post-processor and efficiently compute soft information. Even though the post-processor cannot consider all possible sequences like trellis-based detectors can, we demonstrate for a short target response that our post-processing solution does not experience any loss in performance compared to the optimal maximum <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">a posteriori</i> trellis-based soft detector. The complexity of the computations in the post-processor grows only linearly with the target length, as opposed to the exponential growth in complexity in trellis-based detectors. In this way we can efficiently perform nearly optimal pattern-dependent soft detection in the post-processor for a very long target response without base-line wander compensation.