Abstract
In this paper, the author proposes an interference cancellation precoding method (MB-BD) for multiuser MIMO downlink systems. Maximum beamforming (MB) is used for SNR improvement but just in single-substream transmission. MB-BD is based on maximum beamforming for multi-substream transmission and can improve the received SNR at each antenna of desired users because it uses the sum space of the eigenvectors which correspond to the strong eigenvalues (eigen modes) to transmit data for desired users. Comparing to block diagonalization (BD), MB-BD has 6dB gain if we set constraint of BER at 10 -3 and the sum rate of the proposed method is identical to BD. MB-BD has lower computational complexity than BD because it needs only to do one SVD (or EVD) and one QR decomposition for computing the precoding matrix of each user. We can get about 37.5% reduction of computational complexity at transmitter. To enhance the detection performance and lower computational complexity, we use simplified maximum likelihood (SML) at each receiver. Comparing to original ML, we can get about 33.33% reduction of computational complexity at receiver. © 2012 IEEE.