Research On Robust Adaptive Beamforming Based On LADMM

As the most critical technology in array signal processing,Adaptive Beamforming（ABF）technology has undergone decades of research and development,and has been widely used in radar systems,sonar systems,medical diagnostic systems,earthquake monitoring,radio communications,and astronomical work.application.The main function of the adaptive beamformer is to obtain the corresponding adaptive beamforming weight vector through the relevant adaptive beamforming algorithm according to the received signal of the array,and then suppress and weaken the interference and noise according to the weight vector to obtain the desired signal.Since the development of adaptive beamforming technology,robust adaptive beamforming technology has also been developed,which has achieved a high level of robustness and output performance.However,traditional robust adaptive beamformers are The output performance is improved at the expense of algorithm processing efficiency.To this end,in recent years,researchers have gradually focused their research on improving the efficiency of algorithmic operations.Based on the existing work,this thesis proposes two new robust adaptive beamforming algorithms to achieve the effect of compressing the processing time of the algorithm as much as possible while ensuring the output performance.This thesis first proposes a robust adaptive beamforming algorithm based on the Alternating Direction Method of Multipliers（ADMM）.The algorithm use ADMM to solve the MVDR robust adaptive beamforming mathematical model,and obtain the corresponding optimal weight vector.Due to the excellent convergence and decomposability of ADMM,we can effectively reduce the algorithm complexity of the MVDR robust adaptive beamforming algorithm,shorten the processing time of the algorithm,and improve the efficiency of the algorithm while ensuring the beamforming output performance as much as possible.The algorithm is closer to the practical application level.Then,this thesis also proposes a robust adaptive beamforming algorithm based on the combination of ADMM and deep learning theory（Learned—ADMM,LADMM）.We found that the robust adaptive beamforming algorithm based on ADMM will seriously slow down the processing speed of the algorithm in the high SNR range due to the improvement of the residual error of control convergence at high SNR.In order to solve this problem,the algorithm in this thesis choose to let ADMM combined with the deep learning theory,according to the algorithm structure of ADMM,a RNN network with inner and outer layers is built,and the neural network is trained according to the data set constructed by the traditional MVDR robust adaptive beamforming algorithm to obtain the corresponding model.The input of the model is the noise covariance matrix and the signal covariance matrix,and the output of the second stage is the optimal steering vector.Research shows that the algorithm can effectively solve the problem of serious increase of processing time in the former high SNR interval,and ensure its output performance.