Study of Low Complexity Implementation of Block Diagonalization Precoding

Block diagonalization (BD) algorithm is an efficient precoding technique for multiuser multiple-input multiple-output (MU-MIMO) system. In previous researches, BD precoding was mostly realized based on the singular value decomposition (SVD). However, in the cooperative transmission scenario where the dimension of channel matrix is significantly enlarged compared with the single base station (BS) transmission scenario, implementing SVD decomposition requires heavy computational amount, which is impractical in some real systems. The main contribution of this work is a low complexity solution scheme for BD. When the number of antennas of the receiver is two, instead of using SVD, the proposed scheme computes the null space and eigenvectors by solving linear equations and projection method, respectively. It has a lower complexity when the number of transmit antennas is not large. For example, when the number of transmit antennas is 20, the total computational amounts of the proposed scheme is about 53000 flops while it is around 78000 flops in the SVD based scheme.

[1]  Jungwoo Lee,et al.  Low complexity MIMO scheduling with channel decomposition using capacity upperbound , 2008, IEEE Transactions on Communications.

[2]  Ross D. Murch,et al.  A transmit preprocessing technique for multiuser MIMO systems using a decomposition approach , 2004, IEEE Transactions on Wireless Communications.

[3]  Gene H. Golub,et al.  Matrix computations , 1983 .

[4]  Erik Dahlman,et al.  3G Evolution: HSPA and LTE for Mobile Broadband , 2007 .

[5]  Jeffrey G. Andrews,et al.  Transmit Selection Diversity for Unitary Precoded Multiuser Spatial Multiplexing Systems With Linear Receivers , 2007, IEEE Transactions on Signal Processing.

[6]  Reinaldo A. Valenzuela,et al.  Network coordination for spectrally efficient communications in cellular systems , 2006, IEEE Wireless Communications.

[7]  Gene H. Golub,et al.  Matrix computations (3rd ed.) , 1996 .

[8]  Poras T. Balsara,et al.  VLSI Architecture for Matrix Inversion using Modified Gram-Schmidt based QR Decomposition , 2007, 20th International Conference on VLSI Design held jointly with 6th International Conference on Embedded Systems (VLSID'07).