Fast Prediction of Radio Channel Large-Scale Parameters Considering Environmental Factors

Radio channel modelling is a crucial technique for the development of sixth generation (6G) and beyond 6G (B6G) networks. With the increase in environmental complexity and technical diversity, channel modelling techniques are eventually required to cover environmental information as a digital twin channel (DTC) for complete simulation information, acceptable prediction accuracy, and low modelling complexity. However, current mainstream channel models, the geometry-based stochastic channel models (GSCMs), are rooted in statistical distributions without physical surrounding information, and so they cannot be further evolved into an alternative option for future channel models.

Under these circumstances, in this thesis, we propose a potential method for enhancing current GSCMs with environmental considerations for the future development of channel modelling. We develop the adaptive multiple path loss exponent (AMPLE) framework and its large-scale parameter (LSP) extension (AMPLE-LSP), incorporating environmental information directly into statistical modelling. This framework bridges the gap between deterministic ray-based models and conventional statistical models, achieving significantly improved prediction accuracy while preserving LSP simulation efficiency. Extensively validated via measurements and ray-based simulations, these models consistently outperform widely used models such as those defined by the Third Generation Partnership Project (3GPP) and the Fifth Generation Channel Model Special Interest Group (5GCMSIG), at a comparable computational cost. These results provide a practical and scalable solution for realistic vehicular and wireless network studies, opening a new direction for future GSCMs.