Time-frequency graph enhancement of communication signals based on generative diffusion model in low signal-to-noise ratio environment

Oct 26, 2026

Upload by Bulrr

In low signal-to-noise ratio (LSNR) environments, received communication signals suffer heavy degradation from noise interference, often preventing traditional enhancement methods from improving signal quality. This paper presents a dual-stage signal enhancement approach built on an improved DiffBIR model, using time-frequency diagrams to recover signals adaptively. The Inception-based multi-scale feature extraction supports a PFA pixel fusion attention module for robust noise suppression, while a diffusion generation stage restores high-frequency details via step-by-step conditional denoising. Experiments demonstrate strong performance, especially under very low SNR, while better preserving time-frequency characteristics for communications, radar, and acoustic processing.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages21

Reading Time53 min

AccessFree to download

Abstract

1. Introduction

Low SNR impact on communication signals

Time-frequency diagrams and noise interference

Diffusion models for complex noise environments

Dual-stage improved DiffBIR framework

Xijing University Xijing University Xijing University DiffBIR model, LSNR, Signal recovery, Time-frequency images, Wideband communication signals June 16th, 2025 [https://doi.org/10.21203/rs.3.rs-6746628/v1](https://doi.org/10.21203/rs.3.rs-6746628/v1) 􀁱 􀅏 This work is licensed under a Creative Commons Attribution 4.0 International License. Read Full License No competing interests reported. Time-frequency graph enhancement of communication signals based on generative diffusion model in low signal-to-noise ratio environment Cao wenqing1, Zhu Rui 1, and Chen Yarong1* 1 Xijing University, School of Electronic Information, Xi'an, 710123, China *[corresponding.author@email.example:](corresponding.author@email.example:) [2234134078@qq.com](2234134078@qq.com)[ ](2234134078@qq.com)Abstract In low signal-to-noise ratio (LSNR) environments, the processing of communication signals faces significant noise interference. Traditional signal enhancement methods often perform poorly under these conditions, making it difficult to effectively improve signal quality. To address this challenge, this paper proposes a dual-stage signal enhancement method based on an improved DiffBIR model (Diffusion-based Blind Image Restoration Model), aiming to efficiently enhance received signals through time-frequency diagrams, thereby improving signal quality in low SNR environments. The improved DiffBIR model combines the advantages of deep learning and diffusion processes, utilizing Inception and PFA modules to achieve adaptive signal recovery in the time-frequency domain. The Inception module provides a rich feature foundation for the PFA module through a multi-scale feature extraction mechanism, while the PFA module further enhances the accuracy of signal recovery by optimizing the weight distribution of signal regions. Experimental results show that under low SNR conditions, the proposed improved DiffBIR model significantly outperforms traditional signal enhancement methods, particularly in scenarios with very low SNR, where the enhancement effect is especially pronounced. This method offers an innovative solution for enhancing received signals, not only demonstrating high noise suppression capabilities but also better preserving the time-frequency characteristics of the signal. It has broad application prospects, particularly in fields such as communication, radar, and acoustic signal processing. The code and data supporting this research have been stored on GitHub, with the link being [https://github.com/18291716943/demo](https://github.com/18291716943/demo). Key word:DiffBIR model, LSNR, Signal recovery,Time-frequency images,Wideband communication signals. 1. Introduction In low signal-to-noise ratio (LSNR) environmentsError! Reference source not found. , communication signals are severely affected by noise interference, leading to the masking or loss of signal features, which in turn impacts the transmission quality of communication systems and the reliability of subsequent processing. Particularly, the destruction of highfrequency components and time-frequency characteristics under low SNR conditions makes signal enhancement and restorationError! Reference source not found.Error! Reference source not found. an important research topic. This paper addresses this challenge by proposing a dual-stage signal enhancement method based on the DiffBIR framework, achieving efficient communication signal enhancement through innovative model design and diffusion generation mechanisms. Time-frequency diagrams, as a toolError! Reference source not found. that simultaneously displays the time and frequency distribution of signals, play a crucial role in signal feature extraction and enhancement. However, in low SNR environments, noise interference can blur or even erase key information in time-frequency diagrams. Traditional signal processing methods, while capable of extracting signal features to some extent, often fall short i

FAQs

What problem does the paper address in low SNR communication environments?

In low signal-to-noise ratio (LSNR) environments, noise interference can mask or destroy signal features, especially high-frequency components and time-frequency characteristics, reducing transmission quality and the reliability of downstream processing.

How does the improved DiffBIR model perform dual-stage signal enhancement?

The first stage uses a degradation removal module with multi-scale feature extraction (Inception/LE_RSTB) and a pixel fusion attention module (PFA) to suppress noise while preserving global signal structure. The second stage uses a diffusion generation module with step-by-step conditional denoising to recover time-frequency details and high-frequency content.

What benefits are reported for very low SNR conditions and broader applications?

Experiments indicate the improved model significantly outperforms traditional methods, with especially pronounced enhancement effects in very low SNR scenarios. The approach also aims to better preserve time-frequency characteristics and is suggested as useful for communications, radar, and acoustic signal processing.

Oct 26, 2026

Upload by Bulrr