HA-ESNet: A Hierarchical Attention With Echo State Network-Based Dynamic Low-Complexity Channel Estimation in FSO Communication Links Under Turbulent Channel Conditions

Abstract

In today's rapidly evolving communication landscape, free space optical (FSO) communication systems face significant challenges when operating under atmospheric turbulence conditions. The specific characteristics of gamma–gamma turbulence introduce signal fading, scintillation, and potential link failures, impacting the reliability and performance of data transmission. To ensure high-quality and reliable communication in such challenging environments, there is a critical need for low-complexity parameter estimation techniques with low bit error rate (BER) and mean square error (MSE). Addressing these challenges, this paper proposes a low-complexity channel estimation design named hierarchical attention echo state network (HA-ESNet) model over gamma–gamma turbulence channels in FSO communications. The HA-ESNet model leverages advanced deep learning techniques, attention mechanisms, and the echo state network (ESN) architecture to enhance parameter estimation accuracy and robustness. The hierarchical attention mechanism allows the network to selectively focus on informative channel characteristics while suppressing noise and irrelevant information. This selective attention enables the model to prioritize critical features and adapt to changing channel conditions effectively. The HA-ESNet model's unique architecture combines the benefits of hierarchical attention mechanisms and ESN components to optimize signal transmission, adapt to channel variability, and improve training efficiency. By capturing the nonlinear dynamics of FSO channels through reservoir computing with echo state properties, the HA-ESNet model can effectively model and adapt to the complex turbulence-induced dynamics. Simulation results demonstrate the strong performance of the HA-ESNet model in estimating parameters over turbulent FSO channels. The model achieves low BER, low MSE, and minimal computational complexity, showcasing its robustness and adaptability in capturing the dynamics of turbulent channels. The innovative approach of HA-ESNet significantly enhances the reliability and performance of FSO communication systems in challenging atmospheric conditions, offering a promising solution for improving data transmission in FSO networks.