Deep Learning Aided Parametric Channel Covariance Matrix Estimation for Millimeter Wave Hybrid Massive MIMO

TL;DR

This paper proposes a deep neural network-based method for estimating the channel covariance matrix in millimeter-wave massive MIMO systems, improving computational efficiency and accuracy in high-correlation channels.

Contribution

It introduces a novel deep learning approach for parametric CCM estimation tailored for mmWave massive MIMO, addressing computational and accuracy limitations of existing methods.

Findings

Achieves lower estimation errors compared to traditional methods

Reduces computational complexity in CCM estimation

Effective in high spatial correlation environments

Abstract

Millimeter-wave (mmWave) channels, which occupy frequency ranges much higher than those being used in previous wireless communications systems, are utilized to meet the increased throughput requirements that come with 5G communications. The high levels of attenuation experienced by electromagnetic waves in these frequencies causes MIMO channels to have high spatial correlation. To attain desirable error performances, systems require knowledge about the channel correlations. In this thesis, a deep neural network aided method is proposed for the parametric estimation of the channel covariance matrix (CCM), which contains information regarding the channel correlations. When compared to some methods found in the literature, the proposed method yields satisfactory performance in terms of both computational complexity and channel estimation errors.