Performance Analysis of Deep Learning based Hybrid Precoder/Combiner in Millimeter Wave Massive MIMO Architecture

This paper presents a performance analysis of a Deep Learning based Hybrid Precoder/Combiner for Millimeter Wave (mmWave) Massive Multiple Input and Multiple Output (MIMO) Architecture. Unlike traditional hybrid precoder techniques, which suffer from high energy consumption and underutilization of spectrum, this approach leverages deep learning to address these challenges in mmWave Massive MIMO systems. Specifically, a deep learning model is developed to optimize a cascaded hybrid precoder/combiner using an improved Convolutional Neural Network (CNN). The model predicts the analog precoder/combiner according to channel vector, and then the digital precoder/combiner vector is predicted using the Approximate Message Passing (AMP) algorithm to improve a spectral efficiency of the mmWave Massive MIMO Architecture. The proposed Deep Learning (DL)-based hybrid precoder/combiner using AMP algorithm is provide 15% improvement in achievable rate for the given channel measurement M
t
=M
r
=8 value compared to the existing Algorithm.