# Denoising and Augmentation: A Dual Use of Diffusion Model for Enhanced CSI Recovery

arXiv: https://arxiv.org/abs/2510.02744 · HTML: https://mlsanity.com/paper/2510.02744/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Placed in Diffusion for compression and wireless, one of 14 directions in Diffusion models.
- Builds on 9 of its 15 references that are in the map.
- First posted on arXiv 3 Oct 2025.

## Placed in

- Diffusion models / [Diffusion for compression and wireless](https://mlsanity.com/diffusion/diffusion-for-compression-and-wireless/)

## Builds on (most cited first, 9 of 9)

- [Denoising Diffusion Probabilistic Models](https://mlsanity.com/paper/2006.11239/) (2020)
- [Deep Unsupervised Learning using Nonequilibrium Thermodynamics](https://mlsanity.com/paper/1503.03585/) (2015)
- [Image Super-Resolution Using Deep Convolutional Networks](https://mlsanity.com/paper/1501.00092/) (2014)
- [Deep Learning-Based Channel Estimation](https://mlsanity.com/paper/1810.05893/) (2018)
- [CDDM: Channel Denoising Diffusion Models for Wireless Semantic Communications](https://mlsanity.com/paper/2309.08895/) (2023)
- [At the Dawn of Generative AI Era: A Tutorial-cum-Survey on New Frontiers in 6G Wireless Intelligence](https://mlsanity.com/paper/2402.18587/) (2024)
- [Diffusion-based Generative Prior for Low-Complexity MIMO Channel Estimation](https://mlsanity.com/paper/2403.03545/) (2024)
- [Channel Modeling Aided Dataset Generation for AI-Enabled CSI Feedback: Advances, Challenges, and Solutions](https://mlsanity.com/paper/2407.00896/) (2024)
- [Augmenting Channel Simulator and Semi- Supervised Learning for Efficient Indoor Positioning](https://mlsanity.com/paper/2408.00429/) (2024)

## Built on by (most cited first, 1 of 1)

- [Score-Based Conditional Flow Models for MIMO Receiver Design with Superimposed Pilots](https://mlsanity.com/paper/2602.21654/) (2026)

## Abstract (arXiv)

This letter introduces a dual application of denoising diffusion probabilistic model (DDPM)-based channel estimation algorithm integrating data denoising and augmentation. Denoising addresses the severe noise in raw signals at pilot locations, which can impair channel estimation accuracy. An unsupervised structure is proposed to clean field data without prior knowledge of pure channel information. Data augmentation is crucial due to the data-intensive nature of training deep learning (DL) networks for channel state information (CSI) estimation. The network generates new channel data by adjusting reverse steps, enriching the training dataset. To manage varying signal-to-noise ratios (SNRs) in communication data, a piecewise forward strategy is proposed to enhance the DDPM convergence precision. The link-level simulations indicate that the proposed scheme achieves a superior tradeoff between precision and computational cost compared to existing benchmarks.