遇见数据集

PhysDiff-THz: Generated THz Channel Data and Code for Physics-Embedded Diffusion Model with RIS-Aided Physical Layer Security

收藏
Zenodo2026-06-27 更新2026-06-17 收录
官方服务:

资源简介:

This dataset Contents:1. Generated channel data (NPZ format): - SJTU scenarios: 7 scenarios (corridor, hallway, atrium, LoS/NLoS, 306-371 GHz) - USC D2D scenarios: 2 scenarios (LoS/NLoS, 145 GHz) - TUB measurement scenarios: 5 scenarios (aircraft cabin, conference room, data center, production hall, shopping mall, 300 GHz) - Fused dataset: train/test/val splits 2. Code: - models/: Diffusion model implementation and molecular absorption prior - algorithms/: DiffInit-BF, deep-unfolded WMMSE, baseline algorithms - analysis/: KS fidelity tests, Monte Carlo PLS simulation, imperfect CSI analysis, theorem verification - data/: Data preprocessing scripts (MAT-to-NPZ conversion, dataset fusion) 3. Pre-trained model weights: - physdiff_thz_v1.pt (4.9 MB) Source data references:- SJTU Channel Generator v1.5: https://github.com/zheng-xy/SJTUChannelGenerator- TU Braunschweig 300 GHz measurements: TU Braunschweig Institutional Repository- USC WiDeS D2D model: Contact WiDeS group- ITU-R P.676-13: https://www.itu.int/rec/R-REC-P.676/

提供机构:
Zenodo
创建时间:
2026-06-16
二维码
社区交流群
二维码
科研交流群
商业服务