Research On Sampling Matrix Inverse Beamforming System For Massive MIMO

With the continuous development of mobile communication technology,massive multiple-input multiple-output(Massive MIMO)technology has become one of the most popular communication technologies at present.In traditional MIMO technology,the number of antennas is small,so the beamforming algorithm can be implemented on the serial processor,but in Massive MIMO technology,due to the large number of antennas,the amount of calculation of the beamforming algorithm will also increase accordingly.Increasing,continuing to use serial processors will not be able to meet the real-time requirements of the beamforming system,so this paper considers the use of parallel processors FPGA to accelerate the design of the algorithm.The adaptive beamforming algorithm is the core technology of the adaptive antenna.By adjusting the amplitude and phase of each element of the array antenna,the main direction of the beamformer points to the direction of the desired signal and forms a null in the direction of the interference signal.Classical adaptive beamforming algorithms include Least Mean Square(LMS)algorithm,Sampling Matrix Inversion(SMI)algorithm,etc.Among them,the SMI algorithm belongs to the open-loop algorithm,also known as the direct solution method,which has faster convergence speed,but will be affected by constraints.The direct inversion of the variance matrix is limited by the large amount of calculations,and its beamforming performance will also be affected by the number of sampling snapshots.However,with the continuous upgrading of chip products,the computing power of parallel processors has been continuously improved,which can just make up for the large amount of calculation of the open-loop algorithm.The robustness of the beamformer can also be better controlled by diagonal loading and other methods.Therefore,the SMI algorithm is chosen as the research object of this paper.In this paper,FPGA is used as the parallel processing platform,and CPU is used as the serial processing platform to build a CPU+FPGA parallel heterogeneous SMI algorithm implementation framework to solve the beam instability problem of the SMI beamformer and the beamforming algorithm operation under the Massive MIMO system.The problem of too large amount is solved,and the performance of the algorithm is simulated and evaluated and verified at the board level.The main research content of this paper is as follows:(1)Aiming at the problems of beam distortion and main lobe shift caused by the insufficient number of sampling snapshots of the SMI algorithm in real-time systems,this paper proposes a new genetic algorithm based on variable genetic operators to find the diagonal loading factor Excellent,using the improved genetic algorithm to solve the problem of loading amount selection.Then the perturbation of the small eigenvalues of the covariance matrix is suppressed by applying a diagonal loading method to the sampled covariance matrix,thereby obtaining a stable pattern and lower sidelobes.(2)Aiming at problems such as the numerical instability and poor real-time performance of the algorithm caused by the large amount of calculation and high complexity of the matrix inversion module in the SMI algorithm,this paper uses FPGA to accelerate the calculation part in parallel,and proposes a new algorithm for triangulation The dedicated systolic array structure for matrix inversion and the new block matrix multiplication structure improve the resource utilization and calculation rate of the hardware implementation of matrix inversion,thereby improving the real-time performance of the SMI beamforming algorithm.Compared with the existing robust beamforming algorithm,the diagonal loading factor determination method based on the improved genetic algorithm proposed in this paper has better main beam directivity and stronger interference suppression ability;at the same time,the FPGAbased Compared with the existing state-of-the-art related technologies,the matrix inversion algorithm design achieves the highest frequency and the lowest data delay with the least resources.Therefore,the method proposed in this paper greatly improves the robustness and real-time performance of the beamforming system.