基于小波阈值去噪与时频图像检测的信号调制识别技术
Signal Modulation Recognition Based on Wavelet Threshold Denoising and Time-frequency Image Detection
孙思燕 1张伟雄 1唐娉 2郑柯 2张正2
作者信息
- 1. 中国科学院空天信息创新研究院,北京 100094;中国科学院大学电子电气与通信工程学院,北京 100049
- 2. 中国科学院空天信息创新研究院,北京 100094
- 折叠
摘要
随着通信环境的日益复杂化,信号调制识别变得越来越重要.针对低信噪比下数字信号调制识别困难的问题,提出一种基于小波阈值去噪与时频图像检测的调制识别方法.该方法将接收到的实信号转换成解析信号,通过小波阈值法对解析信号进行去噪处理.引入时频重排技术将去噪后的一维信号转换成二维时频图像,通过双线性插值缩放图像,得到适应网络输入大小的时频图.将时频图输入 VGG网络中训练识别.实验结果显示,提出的调制识别方法对于低信噪比下的调制识别问题表现优异.
Abstract
With the increasing complexity of communication environments,signal modulation recognition has become increasingly important.A modulation recognition method is proposed based on wavelet threshold denoising and time-frequency image detection to address the difficulty of digital signal modulation recognition at low signal-to-noise ratios.The method firstly converts the received real signal into an analytical signal and then denoises the analytical signal by the wavelet threshold method.Then the time-frequency reassignment technology is introduced to convert the denoised one-dimensional signal into a two-dimensional time-frequency image,and bilinear interpolation is used to scale the image to obtain a time-frequency image adapted to the size of the network input.Finally,the time-frequency map is input into the VGG network for training and recognition.The experimental results show that the proposed modulation recognition method performs well for modulation recognition under low signal to noise ratio.
关键词
数字信号调制识别/时频分析/卷积神经网络Key words
digital signal modulation recognition/time-frequency analysis/convolutional neural network引用本文复制引用
出版年
2024
国家科技期刊平台
NSTL
维普
万方数据