现代电子技术2025，Vol.48Issue(19)：63-72,10.DOI:10.16652/j.issn.1004-373x.2025.19.011
基于深度学习的非合作通信信号时频参数估计
Deep learning based time-frequency parameter estimation for non-cooperative communication signals
郑皓羽 1王文彬 2贾哲 3王天乐 4杨峰1
作者信息
- 1. 上海交通大学,上海 200240
- 2. 杭州航海仪器有限公司,浙江 杭州 310024
- 3. 上海交通大学,上海 200240||中国人民解放军93129部队,北京 100000
- 4. 上海交通大学,上海 200240||中国电子科技集团公司第十研究所,四川 成都 610036
- 折叠
摘要
Abstract
A time-frequency parameter estimation framework based on deep learning is proposed.This framework is designed to augment the precision and robustness of parameter estimation for communication signals in non-cooperative contexts,such as military surveillance and signal intelligence collection.The study commences with the selection of the multi-frequency time division multiple access(MF-TDMA)signal model as the subject of investigation.This model facilitates the sharing of the same frequency band among multiple users,optimizing spectral utilization by the dynamic allocation of time-frequency resources.Quantitative simulated samples are generated by utilizing the attributes of MF-TDMA signals,and a dataset is constructed employing short-time Fourier transform(STFT)spectrograms and their enhanced spectral representations.Subsequently,a deep learning framework,based on the YOLOv8 network in the object detection network repertoire,is developed and trained on the custom dataset to determine the location of each burst in the image.The experimental outcomes demonstrate that the incorporation of enhanced spectral data as input elevates detection accuracy markedly and diminishes the estimation offset.Moreover,the proposed deep learning system framework maintains high detection accuracy under various signal-to-noise ratio(SNR)conditions,and the YOLOv8 network demonstrates superior performance in comparison with the other typical networks under diverse conditions.关键词
非合作通信信号/时频参数估计/深度学习/YOLOv8网络/MF-TDMA/增强谱图/目标检测Key words
non-cooperative communication signal/time-frequency parameter estimation/deep learning/YOLOv8 network/MF-TDMA/enhanced spectral/object detection分类
信息技术与安全科学引用本文复制引用
郑皓羽,王文彬,贾哲,王天乐,杨峰..基于深度学习的非合作通信信号时频参数估计[J].现代电子技术,2025,48(19):63-72,10.
