Adaptive, iterative, reduced-rank equalization for MIMO channels

We present an adaptive iterative (turbo) decision feedback equalizer (DFE) for multi-input multi-output (MIMO) channels with intersymbol interference (ISI) and multiple receiver antennas. After initial training, the filters are computed directly from the received data and soft outputs of the MAP decoder according to a least squares (LS) cost criterion. The performance is compared with reduced-rank LS estimation methods, based on the multistage Wiener filter (MSWF). Numerical results show that the reduced-rank turbo DFE provides a substantial performance improvement relative to the full-rank turbo DFE with limited training. In addition, the reduced-rank filters can significantly reduce the computational complexity when the number of filter coefficients is large.