kumoh national institute of technology
Networked Systems Lab.

Rizki Rivai Ginanjar, Jae-Min Lee, and Dong-Seong Kim, "Convolutional Neural Network-Based Channel Matrix Recovery for MIMO Communication", 2019 KICS Summer Conference,pp.1473-1474, June 19-21 2019, Jeju Island, South Korea
By :
Date : 2019-06-24
Views : 293

In this paper, a novel deep convolutional neural network (CNN) architecture for channel matrix recovery in the MIMO communication system is proposed. The proposed deep learning architecture is used to compress the size of the channel state information (CSI) that will be used to calculate the precoding vector. The result shows that by enhancing the system by utilizing the dropout technique, higher accuracy was achieved in the testing stage.