Adaptive stream mapping in MIMO-OFDM with linear precoding

Linear precoding techniques are widely used in emerging MIMO-OFDM standards such as 3GPP LTE and WiMAX. These involve mapping a variable number of streams of transmit data symbols to the transmit antennas using precoding matrices selected from a pre-defined set on the basis of channel state information (CSI) fed back from the receiver. Previous work on these schemes and on selection of precoding matrices has assumed that linear detectors are used, but these cannot exploit the full receive-end diversity when multiple streams are transmitted. This paper presents an adaptive precoding scheme using maximum likelihood (ML) detection with a precoder selection scheme based on minimum BER. It shows that full diversity can be achieved, and that a significant gain is available over adaptive linear precoding using linear detection, over antenna selection, and over spatial multiplexing.