Research And Implementation Of Low Complexity Precoding Algorithm In Multiuser MIMO System

With the development of communication technology,the fifth generation(5G)mobile communication era is coming.As one of the key technologies in 5G technology,Multiuser Multiple Input Multiple Output(MU-MIMO)system introduces interference between users while improving system throughput.The implantation of precoding technology at the base station can suppress the interference between users and improve system performance.It can also reduce the difficulty of client detection and client hardware resource consumption.Therefore,the research of MUMIMO system precoding algorithm has important significance and value.This paper greatly improves Simplified Generalized MMSE Channel Inversion(S-GMI)precoding algorithm and proposes a Simplified Generalized MMSE Channel Inversion Tomlinson-Harashima Precoding(S-GMI-THP)precoding algorithm which is suitable for MU-MIMO communication system.In order to reduce the complexity of the algorithm,the algorithm uses low-complexity QR decomposition to eliminate interference between users.To improve system performance,the algorithm also utilizes high-performance Tomlinson-Harashima Precoding(THP)which eliminates the interference between internal data streams of individual users.In this thesis,the overall hardware architecture of the SGMI-THP precoding algorithm is proposed.And the low-complexity matrix inversion algorithm based on Sherman-Morrison theorem is used to implement the 8×8 scale matrix inversion operation which is the core operation in the precoding algorithm.For the problem that the data in the matrix inversion algorithm grows too fast,this paper proposes the solution at the algorithm level.In the hardware implementation process,the hardware structure of the matrix inversion module is optimized by using the conjugate symmetry property of the inverse matrix to be solved.The simulation result shows that the S-GMI-THP precoding algorithm proposed in this paper has better bit error rate performance than the classic Minimum Mean Square Error(MMSE),S-GMI,THP and Block Diagonalization(BD)algorithm in MU-MIMO system when there are 4 users at the receiving end and each user configures two antennas.And it has a performance improvement of 7.46 dB compared to the S-GMI algorithm and a boost gain of 1.2 dB compared to the MMSE algorithm.The complexity of the algorithm is only 35% of the S-GMI algorithm.For the matrix inversion module which is the core module of the S-GMI-THP algorithm,the structure proposed in this paper has certain advantages in hardware resource consumption and throughput compared with LU decomposition method and QR decomposition method.The FPGA verification result shows that the matrix inversion module used in this thesis achieves a clock frequency of 200 MHz,and reaches the throughput of 8.14 cycles/matrix with 55.7k LUTs resource consumption.