A low complexity S-GMI-THP precoding algorithm and hardware implementation for multi-user MIMO systems

Block Diagonalization precoding technology is widely used in Multiple-Input Multiple-Output system. The computational complexity of the algorithm is mainly derived from two singular value decomposition operations. The Simplified Generalized MMSE Channel Inversion algorithm proposes to replace the singular value decomposition in the BD algorithm with a QR decomposition. This paper proposes the S-GMI-THP algorithm, which uses Tomlinson-Harashima Precoding instead of the singular value decomposition in the S-GMI algorithm, which greatly reduces the complexity. The analysis and simulation results show that the proposed S-GMI-THP algorithm can achieve significant BER performance gain compared to THP and MMSE algorithms. The article also gives the overall block diagram of the S-GMI-THP algorithm and the hardware implementation architecture of the 8×8 scale matrix inversion module, which achieves 23.8M Matrix/s throughput with low resources on the virtex-7 platform.