Study On Nonlinear Algorithm In Hybrid Beamforming System

The increasing demand for data capacity and spectrum efficiency has accelerated the development of fifth-generation communications(5G).MIMO technology combined with appropriate beamforming technology can significantly improve spectrum efficiency.In order to reduce the increasing RF hardware cost,the best solution is to allocate beamforming to the digital domain and the hybrid beamforming structure in the RF circuit.In recently proposed fifth-generation systems,the rapidly increasing demand for RF chains not only increases the size of the transmitter but also creates greater nonlinear distortion,necessitating the use of effective power amplifier linearization technology.This thesis studies the nonlinear algorithm based on hybrid beamforming.The main contents and innovations are as follows:1.The full digital precoding scheme based on SVD singular value decomposition is studied,and its performance is regarded as the optimal value of hybrid beamforming performance,and the beamforming based on orthogonal matching pursuit(OMP)algorithm is studied,using the response matrix of the antenna array Design a hybrid beamforming matrix.On this basis,a shared-based hybrid beamforming algorithm is proposed.The hybrid beamforming matrix at the transmitting end is designed to make it as close as possible to the optimal matrix of all-digital precoding,and then the hybrid beamforming matrix at the receiving end is adjusted according to the hybrid beamforming matrix.The matrix makes the performance of the system reach the best,and its spectral efficiency is closer to full digital precoding than the OMP algorithm.2.In the beamforming-based transmitting mechanism,the traditional digital predistortion method is used to compensate the nonlinear distortion of the power amplifier.It is necessary to construct a separate digital predistorter for each power amplifier,which greatly increases the complexity of the system.To solve this problem,a digital predistortion structure based on beam orientation(BO-DPD)is proposed,which uses the output of all power amplifiers to synthesize the main beam signal,and performs digital predistortion processing on the main beam signal to achieve the purpose of compensating for nonlinearity and effectively reduce system complexity.degree.At the same time,in order to solve the problem that the abovementioned predistortion structure needs to collect the output of all power amplifiers,and the hardware cost of the system is high,a digital predistortion structure based on single power amplifier feedback forward modeling(FM-DPD)is proposed,and the feedback of a single power amplifier is used for modeling.It is estimated that the output signal of the power amplifier is pre-distorted to compensate the nonlinearity of the power amplifier,which greatly reduces the hardware cost of the system.3.Based on the above,an adaptive iterative algorithm is used to learn the parameters of the digital predistorter.In view of the poor convergence performance of the traditional LMS algorithm,the PSO-LMS algorithm combined with particle swarm optimization and the AFSA-LMS algorithm combined with artificial fish school optimization are used for parameter learning,which accelerates the convergence speed of the algorithm while compensating for nonlinear distortions.The optimization algorithm is also suitable for the RLS algorithm.Compared with traditional adaptive algorithms,the algorithm combined with swarm intelligence optimization has a faster convergence rate,and compared with the adaptive algorithm combined with particle swarm optimization,the adaptive algorithm combined with artificial fish swarm optimization has a faster convergence rate.