Beam Division Multiple Access And Omnidirectional Precoding Transmission Method And Implementation In Massive MIMO Systems

The 5th generation mobile communication(5G)is needed to meet the technical requirements including huge capacity,huge connections,ultra low delay,and high speed mobile.To improve the spectral efficiency and power efficiency have became the core issues.The massive MIMO technology allocates large-scale array antennas in the base side and exploits spatial dimension resources,which can significantly improve system spectrum efficiency and power efficiency,but there are problems such as high complexity and excessive pilot cost.The beam division multiple access(BDMA)transmission method utilizes the sparsity of the beam domain channel to decompose the multi-user large-scale MIMO channel into multiple single-user MIMO channels.Then each user communicates with the base station on non-overlapping beams,thus reducing the system complexity and pilot cost.This thesis mainly investigates the BDMA transmission method for users' data in large-scale broadband MIMO system and the omni-directional transmission method of public signals.Meanwhile,the corresponding FPGA implementation and verification for the algorithm are finished in largescale MIMO prototype verification platform.First of all,the channel model of signal transmission for the scenario of massive MIMO broadband system and the BDMA transmission method are investigated.By analyzing wireless channel time-domain signal transmission model in massive MIMO system,we obtain the influence factors of the spatial broadband effect.According to the transmission channel frequency response of OFDM symbols,the new base station antenna steering vectors are obtained.Then the definition of the beam-domain channel in the classical scenario and the four steps of implementing the BDMA method are given,and for large-scale MIMO broadband system,we obtain the beam-domain channel transformation matrix in the broadband scenario.New beamforming matrix brings huge increase of computational cost for beamfield channel transform.In order to solve this problem,this thesis presents a beamforming algorithm based on Chirp-Z transform,which greatly reduces the computational cost.Second,we investigates the omnidirectional precoding for the transmission of public signals represented by the downlink synchronization signals and the time and frequency synchronization algorithm in massive MIMO system.For omnidirectional precoding design,we first decompose the high-dimensional vector into matrix vector product of low emission signal vector and precoding matrix,which alleviates the problem of large-scale pilot overhead existing in the MIMO channel estimation.Two basic design criteria are proposed that must be satisfied to meet the requirement for reliable cell-wide coverage of signal power in angle domain and equal power on each antenna at any instant time.By utilizing ZC sequence and,we give examples of omnidirectional precoding matrix design in uniform linear array,the transmission scheme of the downlink synchronization signal and the corresponding time-frequency synchronization algorithm are proposed,the algorithm performance simulation is given at last.Then,based on the research of beam division multiple access transmission method and omnidirectional precoding,the architecture design and FPGA implementation of the beam former and omnidirectional precoding are completed at the base station of the massive MIMO prototype verification platform.We first introduce the whole hardware platform architecture,including the design of base station prototype verification system and its subsystems of USRP,BEE7 and server.According to the corresponding design process,circuit design and interface design of OFDM demodulation module and beam former module which are responsible for massive parallel data processing are given.After that,we implement the transmitter and receiver by Verilog based on Xilinx Kintex-7 and Virtex-7 series FPGA.At last,the test results and hardware resource consumption are provided.Finally,we present the design of the terminal side of the large-scale MIMO 5G prototype verification platform,and complete the realization of the downlink synchronization algorithm on the terminal side.The first step is to propose the fixed-point design of the top-level module and the input and output of each submodule through simulation,and compares the performance of the fixed-point and floating point algorithms.Then,according to the flow of downlink time-frequency synchronization algorithm,the circuit design,interface design and timing sequence design of each module are completed.The code is also written in Verilog language and implemented on Kintex-7 series FPGA.