Decoding space-time codes with BLAST architectures

We introduce a coding/decoding scheme matched to a "vertical" BLAST architecture; a code has its words evenly split among the transmit antennas. The subcodes so transmitted by each antenna are decoded in sequence to cancel the spatial interference while a final decoding step is performed on the whole code. We also examine the behavior of zero-forcing (ZF) and minimum-mean-square error (MMSE) BLAST by comparing their error probabilities with those resulting from optimum, i.e., maximum-likelihood (ML), processing. Since, with vertical BLAST, ordering of the columns of the channel-gain matrix is crucial, we also study the performance of algorithms intended to find an optimal (or mildly suboptimal) ordering.