DeepTelecom: A Differentiable Multimodal Framework for Channel Twinning in Dynamic Heterogeneous Wireless Environments

Large-scale multimodal data are a key enabler for integrating large generative artificial intelligence models into wireless networks. However, existing wireless datasets offer limited support for heterogeneous environments, particularly in terms of reusable high-fidelity scenes, scalable differentia