Extension of the MIMO Precoder Based on the Minimum Euclidean Distance: A Cross-Form Matrix

Abstract

Under full channel state information at the transmitter side (Tx-CSI), MIMO precoders can be designed by the optimization of many pertinent criteria, like, for example, the maximizing post-processing signal-to-noise ratio (max-SNR or beamforming solution), or the minimizing weighted mean square error between transmit and receive vector-symbols (W-MMSE solution). These solutions decouple the MIMO channel into b parallel independent datastreams. This diagonal structure reduces the complexity of the maximum likelihood (ML) decisions but the diversity order of these schemes is limited. Recently, we proposed a precoder, max-dmjn solution, which optimizes the exact expression of the minimum Euclidean distance and leads to a non diagonal structure allowing to achieve maximum diversity order. However, the result is available only for two transmit datastreams (6 = 2) and BPSK and QPSK modulations. In this paper, we propose a heuristic method to deal with the case b > 2, which provides a suboptimal, but good solution to this general problem. The new precoder, Equal-dm;n (E-dmin), is based on a non diagonal cross-form structure. It significantly enhances the transmit diversity in the eigen-subchannels. We demonstrate that the achieved diversity order is greater than that of precoders with diagonal structure for the same number of datastreams despite a tradeoff between rate and diversity. This design can also ensure quality of service (QoS) by using an adapted power allocation strategy. Performance comparisons show the BER improvement for MIMO and MIMO-OFDM systems.