Research On Precoding Of Millimeter-wave Large-scale MIMO System And FPGA Implementation

In traditional MIMO systems,precoding technology can suppress pilot pollution and intercell interference.However,one antenna in ful y digital precoding algorithm needs to be equipped with a radio frequency link.In massive MIMO systems,the number of antennas is hundreds of thousands,the number of radio frequency links required is too large and the cost is too high.Fully digital precoding algorithm is not suitable for massive MIMO systems.Therefore,hybrid precoding technology emerges as the times require and becomes the focus of research.In this paper,precoding algorithms for mil imeter-wave massive MIMO systems are studied.Firstly,the research background and significance of the subject are elaborated,and the current research status is introduced.Due to the in-depth study of mil imeter wave band,the focus of precoding research begins to shift from ful y digital precoding to hybrid precoding technology.Then the mil imeter-wave massive MIMO system is described.The mil imeterwave massive MIMO system consists of three parts: transmitter,wireless channel and receiver,and the three parts are introduced in detail.On this basis,the basic principles and performance simulation of three commonly used Linear Precoding algorithms are introduced,and the design of analog precoder and digital precoder based on the hybrid precoding algorithm of alternating minimization is elaborated in detail.In the simulation,the performance of the hybrid precoder based on alternating minimization is compared with that of the ful y digital precoder and the traditional OMP algorithm.Through analysis,the hybrid precoder based on alternating minimization can be obtained.The coding algorithm is superior to the traditional OMP algorithm in performance and approximates the fully digital precoder.Another research focus of this paper is the implementation of zero-forcing precoding on FPGA.Zero-forcing precoding algorithm eliminates the interference between users very wel.In order to realize zero-forcing precoding algorithm,for the zero-forcing precoding matrix,because complex matrix inversion operation is involved in the implementation process,matrix decomposition method is used to solve the inverse matrix on the basis of considering the dimension of precoding matrix.Considering the complexity of implementation,the precoding implementation is divided into six modules.The structure and interface of each module are introduced separately.The function of each module is described in detail.At the same time,the matrix decomposition algorithm based on cholesky is introduced.Through this algorithm,the matrix decomposition and matrix inversion are calculated.The design process of matrix decomposition and matrix inversion is described in detail.The complexity and resource consumption of each module are analyzed and calculated.Finally,the functional simulation of each module is validated,and the design is validated by comparing with the simulation data of matlab.Error analysis shows that the design meets expectations.