Deep Learning-based Hybrid Precoding and Combining Designs for Millimeter Wave MIMO Systems

Abstract

In this paper, we apply deep learning-based (DL) approach to solve the hybrid precoding and combining design problem in millimeter wave (mmWave) multiple-input multiple-output (MIMO) systems. After training process, we feed testing data set into neural network (NN) and obtain phases of RF analog precoders and combiners. Given a RF analog precoder, we can acquire baseband precoders by using least square solution and the similar way is applied to RF analog combiner to acquire baseband combiner. As indicated in the simulation results for the evaluated spectral efficiency based on the outputs of DNN, it shows that the performance of our method is competitive.