Work

Deep Learning-Based Cross-Layer Power Allocation for Downlink Cell-Free Massive Multiple-Input–Multiple-Output Video Communication Systems

Year: 2023

Type: article

Abstract: We propose a deep learning-based cross-layer power allocation method for asymmetric cell-free massive MIMO video communication systems. The proposed cross-layer approach considers physical layer channel state information (CSI) and the application layer rate distortion (RD) function, and it aims to enhance video quality in terms of peak signal-to-noise ratio (PSNR). Our study develops a decentralized deep neural network (DNN) model to capture intricate system patterns, enabling accurate and efficient power allocation decisions. The proposed cross-layer approach includes unsupervised and hybrid (supervised/unsupervised) learning models. The numerical results show that the hybrid method achieves convergence with just 50% of the iterations required by the unsupervised learning model and that it achieves a 1 dB gain in PSNR over the baseline physical layer scheme.

Source: Symmetry

DOI: 10.3390/sym15111968

Authors: Wen-Yen Lin, Tin-Hao Chang, Shu-Ming Tseng

Institutions: National Taichung University of Science and Technology, National Taipei University of Technology

Language: English

Topic: Advanced MIMO Systems Optimization

Subfield: Electrical and Electronic Engineering

Field: Engineering

Domain: Physical Sciences

Open Access status: gold

Estimated APC paid: $2,224

Funders: Beijing University of Technology, National Taipei University of Technology, National Science and Technology Council