A Deep Learning Based Mutual Coupling Correction and DOA Estimation Algorithm
Wenwei Fang, Dingke Yu, Xin Wang, Yuzhang Xi, Zhihui Cao, Chunyi Song, Zhiwei Xu
Abstract
The application of deep neural network (DNN) has become popular in satellite communication scenario that requires real-time DOA estimation. However, the DNN for an antenna system trained in the ideal simulation environment will suffer from serious performance degradation when under the effect of mutual coupling. In this paper, we propose a deep learning based algorithm for DOA estimation with the mutual coupling correction. The two-stage algorithm is composed of a Pix2Pix generative adversarial network and a Resnet-50 deep neural network: the Pix2Pix is trained to correct the mutual coupling effect at the first stage, the Resnet-50 is then trained by transfer learning and apples for DOA estimation at the second stage. The simulation results verify the effectiveness of the proposed algorithm.