Advanced Radio Wave Propagation Models for Next Generation Wireless Networks Using AIML

Exact radio wave propagation modeling is an essential need of designing and optimising the fifth-generation (5 G) and beyond wireless networks in the challenging urban, indoor and high-frequency deployment cases. Although classical propagation models are computationally efficient, they do not provide a good combination of aspects of environmental variability, multipath propagation and frequency-dependent attenuation which occurs at millimeter-wave (mmWave) and terahertz (THz) frequencies. This study aims to eliminate such drawbacks by introducing a superior hybrid radio wave propagation model that combines deterministic ray tracing, statistical channel, and machine learning-based correction that is obtained through empirical means. The proposed model is tested on sub-6 GHz, mmWave and THz frequency bands, in both combination of standardized channel datasets and large-scale real world measurement campaigns in urban, indoor, and rural settings. Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and coefficient of determination$\mathrm{R}^{\mathrm{2}}$are all used to analytically compare performance with traditional models. This has been confirmed in experimental results where the hybrid method yields an average improvement of$22-35 \%$on the prediction accuracy as compared to traditional propagation models, and in particularly dense city and indoor scenarios, large improvements are recorded. The results indicate that hybrid propagation modeling is a powerful and scalable solution to the planning of the next-generation wireless network, spectrum allocation, and performance optimization in 5 G and upcoming 6G networks.