Channel Modeling via Virtual Environment for MIMO–OFDM Systems

Wireless channel modeling plays a crucial role in characterizing the transmission environment and the design of wireless communication networks. However, classical channel modeling methods face significant limitations in accuracy and computational efficiency under the complex and diverse scenarios of the fifth-generation (5G) and sixth-generation (6G) networks. This letter proposes a virtual transmitter (vT)-based channel modeling method for multiple-input-multiple-output orthogonal frequency division multiplexing (MIMO-OFDM) systems that achieves high accuracy with low complexity. By transforming each propagation path into a virtual line-of-sight (LoS) path, a virtual environment has been constructed that eliminates the need for detailed material properties. A scalable vT-based channel modeling neural network (vTCM-Net) is developed to learn its complex field transformations and virtual characteristics. Furthermore, scaling law is derived that quantifies the performance improvement of vTCM-Net with increasing model parameter scales, providing a guide for efficiency-accuracy trade-offs in resource-constrained systems.