Design And Implementation Of Digital-analog Hybrid Precoding In Large-scale MIMO System

With the rapid development of mobile Internet,more and more application scenarios have emerged.It is urgent to develop the fifth generation mobile communication system(5G)to meet people's growing wireless communication needs.Large-scale Multiple-Input Multiple-Output(MIMO)has become one of the key technologies of 5G,which can significantly improve the throughput and spectrum utilization of communication systems.However,in the implementation and application of large-scale MIMO system,there are still some problems such as the surge of data calculation,the increased radio frequency(RF)link overhead and CSI acquisition difficulty.To solve these problems,this thesis designs a large-scale MIMO system architecture with high resource utilization,verifies the prototype,and studies the hybrid digital and analog precoding technology to reduce the RF link overhead.Among them,the large-scale MIMO hybrid digital and analog precoding in dynamic scenes is studied,and a hybrid digital and analog precoding algorithm which can sense the information of environmental change is proposed.The main work of this thesis is as follows:Firstly,according to the characteristics of NI-PXIe hardware platform and the design target of large-scale MIMO system,the hardware resources were analyzed and reasonably distributed.Then,a large-scale MIMO parallel architecture with high resource utilization was designed.This architecture adopts the design concept of parallel processing to solve the problems of high clock rate and large amount of data calculation.At the same time,a more stable scheme of information exchange between boards is proposed,so that the data length transmitted between different FPGAs can be changed adaptively.This scheme ensures that the information exchange rate between boards does not exceed the upper limit of hardware support.This architecture effectively improves the transmission rate of the system and supports the flexible expansion of a larger number of antennas.Then,based on the large-scale MIMO parallel architecture,an ultra-high-speed MIMO communication system was implemented on the NI-PXIe platform,which was able to support error-free real-time transmission of high-definition video streams.And the system throughput reached over 1 Gbps,verifying the feasibility and effectiveness of the architecture.In the specific hardware implementation process,some key FPGA implementation schemes with low cost is proposed.The baseband modules implemented in the system mainly include channel coding/decoding,modulation/demodulation,inverse Fourier transform(IFFT)/Fourier transform(FFT),signal synchronization,channel estimation and equalization,etc.Next,in order to further reduce the hardware cost and power consumption of large-scale MIMO system,the digital-analog hybrid precoding technology is studied.Different hybrid precoding methods are simulated and analyzed in a single user large-scale MIMO system.Considering the performance and complexity of implementation,a hybrid precoding method based on Orthogonal Matching Pursu(OMP)was selected for implementation.And the realized hardware module was coordinated with MATLAB to verify the correctness of the function of the module.Finally,a hybrid digital and analog precoding algorithm based on machine learning is proposed for large-scale MIMO communication systems in dynamic scenarios.The algorithm can perceive the change of environment and learn the potential probability information needed,and then adaptively adjust the solution strategy of analog precoding matrix.At the same time,this hybrid precoding algorithm does not need to know CSI in advance,avoids the difficulty of CSI acquisition in large-scale MIMO.And the algorithm is suitable for all kinds of complex communication scenarios.Through simulation and comparative analysis,it is found that with the increase of signal-to-noise ratio and antenna number,the performance of the proposed algorithm is gradually close to that of traditional pure digital precoding,which verifies the effectiveness and superiority of the performance of the algorithm.