Suboptimal multi-user MLSE with partial interference cancellation by tentative decisions

Addressing the problem of intersymbol and interchannel interference (ICI and ISI), optimal and reduced complexity maximum-likelihood receivers for multi-user additive white gaussian noise matrix channels are considered. In general, a multi-dimensional MLSE, e.g., a Viterbi-algorithm is unfeasible for severe ICI, ISI and a large number of users. Applying a perfect decorrelating (noise whitening) matrix filter, the triangular structure of the resulting channel matrix allows a simplified ICI cancelling detector. More practically, an ICI cancelling structure without the necessity of a decorrelating matrix is presented. It comprises user-individual one-dimensional ISI-MLSEs where ICI is cancelled out by reliable tentative symbols of neighboring detectors. To overcome the problem of non-reliable tentative decisions of the ISI-MLSEs, they are delivered by an ICI sequence estimator with implemented ISI cancellation. Applying filters for minimum phase impulse responses, sequential decoding algorithms like the M-algorithm are suitable for a reduced complexity joint detection receivers