Channel estimation of massive MIMO UV communication with deep pyramid convolution blind denoising network

Abstract

This paper presents a novel deep learning-based approach for channel estimation in ultraviolet (UV) MIMO systems using a Pyramid Convolutional Blind Denoising Network (CBDNet). The proposed architecture consists of a noise estimation subnetwork and a noise removal subnetwork to effectively handle complex, spatially varying noise in turbulent UV channels. By leveraging multi-scale feature extraction and residual learning, the model accurately estimates the channel matrix under low SNR conditions. Simulation results show that the proposed method achieved the MSE $4.5\times 10^{-4}$ at 20 dB SNR and outperforms recent ML-based techniques, making it highly suitable for practical UV MIMO communication systems.