GNN-Based Super-Resolution for Multipath Channel Generation in Railways

Accurate modeling of wireless channels is crucial for the design and optimization of railway wireless communication systems. Ray-tracing (RT) is a widely used technique for generating multipath channel characteristics, but the high computational resource requirements make it difficult to meet real-time demands. To address this issue, a super-resolution (SR) model based on the Graph Neural Network (GNN) for generating multipath channel data is proposed. The model leverages the ability of GNNs to process graph-structured data, mapping the multipath propagation process to the graph representation. RT generates low-resolution multipath data, which is then improved by the GNN-based SR model to produce high-resolution channel characteristics. Experimental results in the railway scenario demonstrate that the proposed method significantly reduces computation time while maintaining high accuracy. The GNNbased SR model performs excellently in power, spatial, and angular domains, providing a more efficient and accurate solution for railway wireless communication systems.