Attenuation Prediction for 6G Terahertz Satellite-Ground Links: A Review of Mechanisms, Models and Future Directions

Abstract

Terahertz communication has become a key enabling technology for 6G space-air-ground integrated networks, offering ultra-wide bandwidth and Tbps-level transmission capacity to meet the demands of global seamless coverage. However, the propagation of terahertz signals is highly susceptible to the atmospheric boundary layer, where absorption by water vapor/oxygen, scattering by aerosols, and rainfall-induced attenuation collectively cause severe degradation of the satellite-ground link. It has become a central research focus in leading international conferences. This paper systematically reviews the state-of-the-art in terahertz attenuation prediction and classifies existing models into three categories: classical empirical models (e.g., ITU-R P.676), physics-based models (e.g., radiative transfer models), and machine learning-based models (e.g., LSTM and random forests). It comprehensively compares their computational efficiency, prediction accuracy, and applicability across scenarios (e.g., dry vs. humid regions, low vs. high terahertz bands). Finally, this paper identifies key challenges, including scarce high-quality measured data and poor cross-regional adaptability and proposes future directions such as adaptive model design and multi-source data fusion. By delineating the full research context and key development trends in this field, this review delivers a structured, up-to-date reference for fellow researchers and engineering practitioners.