DeepTelecom
收藏资源简介:
DeepTelecom是一个三维数字孪生信道数据集,由浙江大学等研究机构创建。该数据集通过大语言模型辅助构建具有可分割材料-参数化表面的室外和室内场景,并基于Sionna的射线追踪引擎模拟全无线电波传播效果。DeepTelecom利用GPU加速,实时输出同步的多视图图像、信道张量和多尺度衰落轨迹,为无线人工智能研究提供了统一的基准,并为未来通信中的基础模型提供了丰富的训练基础。
DeepTelecom is a 3D digital twin channel dataset created by research institutions including Zhejiang University. This dataset constructs outdoor and indoor scenes with separable material-parametrized surfaces with the assistance of large language models (LLMs), and simulates full radio wave propagation effects using Sionna's ray-tracing engine. Leveraging GPU acceleration, DeepTelecom simultaneously outputs synchronized multi-view images, channel tensors, and multi-scale fading trajectories in real time. It provides a unified benchmark for wireless artificial intelligence research and a rich training foundation for foundation models in future communications.



