benediktfesl/Diffusion_channel_est：论文《基于扩散的生成先验用于低复杂度MIMO通道估计》的源代

项目介绍

benediktfesl/Diffusion_channel_est 是一个托管在 GitHub 上的开源项目，主要使用 Python 编写，采用 BSD 3-Clause "New" or "Revised" License 许可证，目前获得 89 个 Star、13 次 Fork。

仓库全名: benediktfesl/Diffusion_channel_est

README 摘要

Diffusion-Based Channel Estimation Source code of the paper B. Fesl, M. Baur, F. Strasser, M. Joham, and W. Utschick, "Diffusion-Based Generative Prior for Low-Complexity MIMO Channel Estimation," in IEEE Wireless Communications Letters, 2024. [IEEE] [arXiv] Abstract This letter proposes a novel channel estimator based on diffusion models (DMs), one of the currently top-rated generative models, with provable…

完整内容见 GitHub 原始 README 或本页。

常见问题

Diffusion_channel_est 用什么语言开发？

主要使用 Python 编写，采用 BSD 3-Clause "New" or "Revised" License 许可证，目前有 89 个 Star。

Diffusion_channel_est 可以免费商用吗？

如何获取 Diffusion_channel_est 的源码？

可以直接克隆仓库：git clone https://github.com/benediktfesl/Diffusion_channel_est.git，默认分支为 master；安装与运行步骤见仓库 README。

Diffusion_channel_est 最近还在维护吗？

仓库最近一次更新为 6月18日，当前没有未关闭的 Issue；Star 数为 89，可作为社区活跃度的参考。

关于

GitHub 原始描述（英文）

Source code of the Paper "Diffusion-Based Generative Prior for Low-Complexity MIMO Channel Estimation"

channel-estimation diffusion-models generative-model machine-learning mimo

BSD 3-Clause "New" or "Revised" License
89 Stars
13 Forks
1 Watching

描述取自 GitHub 官方 API，保持仓库作者书写的原文。

简要分析

热度分

116

价值分

活跃状态

活跃

GitHub

语言占比

Python

Release

Diffusion-based Channel Estimation WCL 20245月1日

贡献者

benediktfesl

Diffusion-Based Channel Estimation

Source code of the paper

B. Fesl, M. Baur, F. Strasser, M. Joham, and W. Utschick, “Diffusion-Based Generative Prior for Low-Complexity MIMO Channel Estimation,” in IEEE Wireless Communications Letters, 2024.

[IEEE] [arXiv]

Abstract

This letter proposes a novel channel estimator based on diffusion models (DMs), one of the currently top-rated generative models, with provable convergence to the mean square error (MSE)-optimal estimator. A lightweight convolutional neural network (CNN) with positional embedding of the signal-to-noise ratio (SNR) information is designed to learn the channel distribution in the sparse angular domain. Combined with an estimation strategy that avoids stochastic resampling and truncates reverse diffusion steps that account for lower SNR than the given pilot observation, the resulting DM estimator unifies low complexity and memory overhead. Numerical results exhibit better performance than state-of-the-art estimators.

DMSE Scheduler Package

A standalone implementation of the DMSE scheduler is available as a separate package:

pip install diffusers-dmse

[PyPI] [GitHub]

This repository contains an application-specific implementation. The standalone package provides a reusable and diffusers-compatible version of the scheduler.

Requirements

The code is tested with Python 3.10 and Pytorch 2.1.1. For further details, see environment.yml.

Instructions

Download the channel dataset from Zenodo:
https://zenodo.org/records/20737830
Extract the downloaded archive and move the extracted folder bin into the repository root.
To evaluate the pre-trained models used for the plots in the paper, run:

python load_and_eval_dm.py -d cuda:0

To train a DM from scratch and evaluate the performance afterward, run:

python diff_cnn.py -d cuda:0

To evaluate the baseline estimators, run:

python baselines.py

暂无评论

Diffusion_channel_est