Cut-off rate based transmit optimization for spatial multiplexing on general MIMO channels

The use of spatial multiplexing (SM) in multiple-input multiple-output (MIMO) wireless systems promises a linear (in the minimum of the number of transmit and receive antennas) increase in data rate. In practice, the performance of SM depends critically on a variety of channel conditions, including antenna height and spacing, polarization of antennas, and richness of scattering. Transmit correlation has been shown to be detrimental to the performance of SM, since it leads to the existence of preferred spatial directions. In addition, the presence of an ill-conditioned fixed (possibly line-of-sight) component in the channel can severely degrade performance. We present a simple transmit optimization strategy to mitigate partially the impact of unfavorable channel statistics on the performance of SM. The proposed strategy takes the scalar symbol constellation and the channel statistics into account and relies on simple phase-shifting of the multiplexed symbol streams at the transmitter. The phase shifts are chosen such that the cut-off rate of the effective channel (physical channel in combination with finite constellation and ML decoding) is maximized. We find SNR gains of up to 4 dB over the case when no transmit optimization is employed.