Research On Precoding Based On Vector Modulus Algorithm In Massive MIMO System

Massive (Multiple-In Multiple-Out) MIMO as one of the 5G technology has been widely concerned. In Massive MIMO system, precoding is needed to mitigate the channel noise and interference between users. As the channel matrix dimension increasing, the complexity of the precoding algorithm can be greatly increased, low complexity precoding is still been looking for, but the system performance of most of the low complexity precoding algorithm is not good. At the same time, with the increase of users in Massive MIMO system, the interference between users will be increased, which leads to the reduction of the bite error rate performance of the system. In this thesis, the low complexity precoding algorithm will be researched to improve the bit error rate performance of the system. In addition, the channel state estimation according to the configuration of the antenna system needed to know by the signal before transmitted, so before precoding making a choice on the existing antenna, it not only can reduce the radio frequency module of the Massive MIMO system, but can enhance the bit error rate performance of the precoding algorithm.Firstly, the Massive MIMO system model is discussed in this thesis, then in the Rayleigh fading channel model the linear zero forcing (ZF) precoding, minimum mean square error (MMSE) precoding and non-linear dirty paper code(DPC), Tomlinson Harashima (TH) precoding algorithm are compared.With the research of the vector norm descend search algorithm based on vector perturbation theory and the focus discussion of the approximation matrix inverse precoding(AMIP) algorithm based on the Neumann series expansion, the approximation matrix inverse precoding is improved by using the norm descend search algorithm in this thesis. The approximation matrix inverse precoding algorithm could reduce the complexity of the precoding matrix hardware implementation compared with the traditional ZF and MMSE precoding algorithm, but the bite error rate performance is declined. In order to improve the bite error rate performance of the approximate matrix inverse precoding algorithm, the vector perturbation theory is used to perturbation the initial transmitted signal, then the vector norm descent search algorithm is used to obtain the minimum modulus value of multiplicative product of the initial precoding matrix and the disturbance signal to realize the improvement of approximation matrix inverse precoding algorithm. The simulation results show that the bite error rate performance of improved approximate matrix inverse precoding algorithm has improved.In addition, in order to further enhance the bit error rate performance of the improved approximation inverse matrix precoding(NAMIP) algorithm, taking the minimum system error theory into consideration, the channel matrix column vector modulus selection algorithm is used to select the transmit antennas, and then the precoding matrix is realized. The transmit antennas are selected to reduce to antenna radio frequency module of the Massive MIMO system. At the same time, simulation results show that the bite error rate performance of the improved approximation matrix inverse precoding algorithm using antenna selection also has improved.