This repository contains an implementation of using generative diffusion models for high-dimensional channel estimation.
Paper: Generative Diffusion Models for High Dimensional Channel Estimation, in IEEE TWC 2025.
The QuaDRiGa channel dataset used in this repository can be downloaded from download url
It is highly recommended to use a virtual environment (e.g., Anaconda or venv) to manage your dependencies. (Note: If a requirements.txt is provided in future updates, you can simply run pip install -r requirements.txt)
Download the dataset from theprovided Google Drive link. Extract the downloaded files and place them into the data/ directory so that loaders.py can load them correctly.
Run train_diffusion_cnn.py, and the model will be saved in the model/ directory.
To evaluate the channel estimation performance using a trained model, run the testing script (test_diffusion_cnn.py). It will load the test dataset and the pre-trained weights.
[1] X. Zhou, L. Liang, J. Zhang, P. Jiang, Y. Li, and S. Jin, “Generative diffusion models for high dimensional channel estimation,” IEEE Trans. Wireless Commun., vol. 24, no. 7, pp. 5840–5854, Jul. 2025.
[2] B. Fesl, M. Baur, F. Strasser, M. Joham, and W. Utschick, “Diffusion-based generative prior for low-complexity MIMO channel estimation,” Mar. 2024. [Online] Available: http://arxiv.org/abs/2403.03545. (Repository: url)
[3] M. Arvinte and J. I. Tamir, “MIMO channel estimation using score-based generative models,” IEEE Trans. Wireless Commun., vol. 22, no. 6, pp. 3698–3713, Jun. 2023. (Repository: url)
If you have any questions or comments about this work, please feel free to contact xy_zhou@seu.edu.cn.