Newswise — As the underlying foundation of digital twin networks (DTNs), digital twin channels (DTCs) can accurately depict electromagnetic wave propagation in the air interface, thereby supporting DTN-based 6G wireless networks. However, existing methods for constructing DTCs face challenges: the environmental information input to neural networks has high dimensionality, and the correlation between the environment and channels is unclear, leading to a highly complex relationship construction process.

To address this issue, a team of researchers from the State Key Laboratory of Networking and Switching Technology at Beijing University of Posts and Telecommunications and China Mobile Research Institute conducted a study titled “Electromagnetic Wave Property Inspired Radio Environment Knowledge Construction and Artificial Intelligence Based Verification for 6G Digital Twin Channel”.

This research proposes three core solutions:

1. Unified Radio Environment Knowledge (REK) Construction Method: Inspired by electromagnetic wave properties, this method quantifies the propagation contributions of reflection, diffraction, and blockage using easily obtainable three-dimensional coordinates or real-time updated location information, applicable to scenarios with different blockage levels.
2. Effective Scatterer Determination Scheme Based on Random Geometry: This scheme significantly reduces environmental redundancy. It achieves a scatterer selection accuracy of at least 90% and reduces redundancy by 90%, 87%, and 81% in fully open, impending blockage, and fully blocked scenarios, respectively.
3. Lightweight Path Loss Prediction Method: Using a simple two-layer convolutional neural network (CNN), this method takes the REK spectrum as input. It reduces input data dimensionality while clarifying deep relationships between the environment and channels, requiring only 4 ms of testing time with a prediction error of 0.3, effectively lowering network complexity.

Extensive validation was conducted on the BUPTCMCC-DataAI-6G dataset (a 646 m × 290 m simulation scene with 7320 user terminals). Compared with unprocessed location data-based and environmental feature-based methods, the proposed REK-based method improves prediction accuracy by 29.4% and 27.5%, respectively, and reduces time complexity by five orders of magnitude compared to ray tracing-based channel parameter generation.

The paper “Electromagnetic wave property inspired radio environment knowledge construction and artificial intelligence based verification for 6G digital twin channel” is authored by Jialin WANG, Jianhua ZHANG, Yutong SUN, Yuxiang ZHANG, Tao JIANG, and Liang XIA. Full text of the open access paper: https://doi.org/10.1631/FITEE.2400464.