| 摘要: |
| 阵列天线接收到的期望信号和干扰信号，其入射的到达角度(Angle of Arrival，AOA)总是快速变化的，而传统波束形成算法计算量大，无法实时计算。针对这一问题，提出了一种基于深度神经网络的自适应波束形成(Deep Neural Network Adaptive Beamforming，DNNABF)算法，用入射信号AOA组成的向量作为网络输入，网络输出逼近最小方差无失真响应(Minimum Variance Distortionless Response,MVDR)算法求得的权矢量。仿真结果表明，卷积神经网络(Convolutional Neural Network,CNN)与DNNABF方法都能准确拟合MVDR算法权矢量，可在入射信号AOA快速变化时自适应地形成波束和零陷，但DNN计算速度相对MVDR有将近6.5倍的提升，训练模型时间也远低于CNN。 |
| 关键词: 自适应波束形成 到达角度 深度神经网络 |
| DOI： |
|
| 基金项目: |
|
| An adaptive beamforming algorithm based on deep neural network |
| REN Yanfei,DU Ying,ZHANG Jindong |
| (Southwest China Institute of Electronic Technology, Chengdu 610036, China;School of Electroinc Information Engineering,Nanjing University of Aeronautics and Astronautics,Nanjing 211106,China) |
| Abstract: |
| The incoming angles of arrival(AOA) of desired signal and interference signal received by array antenna vary fast,but traditional beamforming algorithms cannot compute in real time due to a large amount of calculation.In order to solve the problem,a deep neural network adaptive beamforming(DNNABF) algorithm is proposed.The vector composed of incoming signals’ AOA is put into the network,and the DNN is trained to approximate to the weight vector of the Minimum Variance Distortionless Response(MVDR) algorithm.Simulation results show that the Convolutional Neural Network（CNN） and DNNABF method can both fit the weight vector of MVDR algorithm accurately and form a beam and nulls adaptively when incoming signals’ AOA vary fast,but the DNN calculates about 6.5 times faster compared with MVDR algorithm,while its training time is much less than CNN. |
| Key words: adaptive beamforming angle of arrival deep neural network |
