Lattice-reduction-aided MMSE precoding for correlated MIMO channels and performance analysis

The lattice-reduction (LR) has been developed to improve the performance of the zero-forcing (ZF) precoder in multiple input multiple output (MIMO) systems. Under the assumptions of uncorrelated flat fading channel model and perfect channel state information at the transmitter (CSIT), an LR-aided ZF precoder is able to collect the full transmit diversity. With the complex Lenstra-Lenstra-Lov´asz (LLL) algorithm and limited feedforward structure, an LR-aided linear minimum-mean-square-error (LMMSE) precoder for spatial correlated MIMO channels and imperfect CSIT is proposed to achieve lower bit error rate (BER). Assuming a time division duplexing (TDD) MIMO system, correlated block flat fading channel and LMMSE uplink channel estimator, it is proved that the proposed LR-aided LMMSE precoder can also obtain the full transmit diversity through an analytical approach. Furthermore, the simulation results show that with the quadrature phase shift keying (QPSK) modulation at the transmitter, the uncoded and coded BERs of the LR-aided LMMSE precoder are lower than that of the traditional LMMSE precoder respectively when Eb/N0 is greater than 10 dB and 12 dB at all correlation coefficients.