Two-dimensional Direction Estimation of Generative Adversarial Beamforming
- Resource Type
- Conference
- Authors
- Liu, Jingben; Dong, Jinxin; Liu, Jiantao; Guo, Lianghao; Yan, Chao
- Source
- 2021 OES China Ocean Acoustics (COA) Ocean Acoustics (COA), 2021 OES China. :821-824 Jul, 2021
- Subject
- Communication, Networking and Broadcast Technologies
Components, Circuits, Devices and Systems
Computing and Processing
Robotics and Control Systems
Signal Processing and Analysis
Training
Machine learning algorithms
Deconvolution
Array signal processing
Oceans
Estimation
Machine learning
two-dimensional beamforming
machine learning
direction estimation
robust
- Language
To solve the problem of large main lobe width of conventional beamforming methods, a two-dimensional deconvolution beamforming method based on machine learning is proposed. The algorithm uses the conditional generative adversarial network to learn the mapping relationship between beam power spectrum and point source distribution. In the process of algorithm training, the two-dimensional scanning results of conventional beamforming are input. The main lobe of the network output is narrower. Since this method inherits the advantages of the conventional beamforming, it is more robust than the adaptive beamforming method in the case of steering vector errors.