Performance Analysis of Deep-Learning Based Symbol Estimation for Image Transmission over a Cooperative System

This paper investigates the impact of an amplify-and-forward (AF) cooperative communication system on image transmission, employing a novel deep learning (DL)-based symbol estimator at the destination terminal (DT) replacing conventional maximum likelihood (ML) detection and performing image denoising via median filtering. Comprehensive analysis and performance comparison of all scenarios in terms of bit error rate (BER) and image quality metrics, including peak signal-to-noise ratio (PSNR), structural similarity index (SSIM), and mean-squared error (MSE) are realized. Our simulations and subsequent analyses demonstrate that DL-based symbol estimation in a cooperative scheme exhibits robust symbol detection and denoising performance for image transmission, comparable to conventional methods, and even outperforms under certain conditions.