Efficient 3D Electromagnetic Environment Reconstruction with Fused Data for 6G Digital Twin Online Channel Modeling

With the development of sixth-generation (6G) wireless communication networks, digital twin online channel models (DTOCM) have been proposed to facilitate efficient 6G network optimization. The reconstruction of a three-dimensional (3D) electromagnetic (EM) environment is a critical step in DTOCM, which aids in the accurate characterization of wireless channels. We propose an efficient method for the reconstruction of the 3D twin EM environment in DTOCM, utilizing joint deep learning and image processing with fused data. First, the building environment is reconstructed using digital maps and triangular meshes to obtain precise 3D building models. Next, deep learning networks are used to perform environmental perception from satellite maps, capturing geospatial information and EM parameters. Finally, these models are integrated, and the ray-tracing (RT) method simulates channel characteristics. We evaluate the proposed method using digital and satellite maps of Meishan City, achieving a comprehensive 3D EM environment reconstruction. The RT simulation results based on 3D EM reconstruction closely align with channel measurement data, demonstrating higher accuracy than the 3D building model.