A Per-User Successive MMSE Precoding Technique in Multiuser MIMO Systems

We propose a per-user successive minimum mean-squared error (PU-SMMSE) precoding technique suitable for the use in multiuser multiple-input multiple-output (MU-MIMO) downlink systems. Its preceding complexity is much simpler as compared with that of the successive MMSE (SMMSE) technique (V. Stankovic and M. Haardt, 2004), without any performance loss. The new technique also takes both the high capacity benefit achievable with MIMO multiplexing and the benefit of space division multiple access (SDMA), assuming channel state information (CSI) at both the transmitter and receiver sides. In the new technique, the columns of the precoding matrix are obtained by computing successively reduced-rank MMSE precoding solutions on a per-user basis. Then, a per-user virtual channel which consists of the per-user channel through precoding is decomposed into orthogonal transmission modes. Finally, a proper power assignment can be applied to the decomposed modes according to some assignment policy. We compare the proposed technique with the SMMSE technique in terms of the bit error rate (BER) performance. We also analyze the computational complexity of the PU-SMMSE technique in comparison with the SMMSE and BD techniques. From computer simulations and complexity analyses, we see that the PU-SMMSE and SMMSE techniques have the same BER performance in both cases of dominant mode transmission (DMT) and multiple mode transmission (MMT). However, the computational complexity of the proposed technique is much lower than that of the SMMSE technique, and also is lower than that of the simple block diagonalization (BD) technique.