Maximum a posteriori decision feedback detection

A new (nonadaptive) detection scheme for the equalisation of uncoded binary data transmitted on noisy, dispersive communications channels is presented. The authors use a simplification of the optimal symbol-by-symbol maximum a posteriori (MAP) estimation criterion, teamed with block processing and decision feedback. The resulting device, the MAP decision feedback detector, does not rely on the assumption of correct state feedback (unlike its predecessor, the block decision feedback equaliser or ´block DFE´), but uses knowledge of certain operational error probabilities to give improved performance over the block DFE. One advantage of the technique over conventional MAP detectors is that channels with long impulse responses can be handled by the incorporation of decision feedback. This has the effect of reducing the dimension of the space on which decisions are formed. They briefly describe some related but simpler suboptimal equaliser structures which are special cases of their detector. They compare the operation of the new detector, the block DFE, the decision feedback equaliser and the Viterbi algorithm on a second order channel. They also include examples of the decision regions formed by the new detector and the block DFE.<>