Hybrid Forward-Backward Ray Tube Propagation Model for Deterministic Multipath Channel Prediction

Deterministic propagation prediction faces a fundamental trade-off: the image method is exact but single-mechanism, while the shooting and bouncing ray method is flexible yet suffers from path omissions due to discrete sampling. This paper proposes the hybrid forward–backward ray tube propagation model (HFB-RTPM), which decouples the exhaustive forward construction of a ray tube tree from the backward path screening at the receiver, handling reflection, refraction, diffraction, and diffuse scattering within a unified framework. A quasi-3D extension and a dual-lobe diffuse scattering model are further introduced. Simulations show 99.8% reflection path completeness (RMSE 0.05 dB) and 100% diffraction completeness under all conditions. With diffuse scattering, the mmWave NLoS RMSE drops from 23.9/16.2 dB to 6.9/6.8 dB at 28/73 GHz, with a single model configuration valid across both frequency bands. In field measurements at 2.3–5.9 GHz across dense urban blocks, the simulation-to-measurement RMSE ranges from 5.6 to 8.5 dB. For large-scale coverage, HFB-RTPM is two orders of magnitude faster than the image method. The proposed method achieves accuracy, efficiency, and physical completeness for deterministic multipath prediction in 6G wireless networks.