基于特征分布密度图的通信调制信号识别
Communication modulation signal recognition based on feature distribution density map
刘静 1李靖超 1林军2
作者信息
- 1. 上海电机学院电子信息学院,上海 201306
- 2. 国网浙江武义县供电公司 电力调度控制中心,浙江金华 321200
- 折叠
摘要
针对直接利用时域特征对信号进行识别存在特征不明显、鲁棒性较差的问题,提出了一种基于特征分布密度图的通信调制信号识别方法.通过对通信调制信号进行采样和分析,提取信号的平均值和偏度,并利用高斯核函数计算点概率密度,生成特征分布密度图,再利用深度学习进行分类识别,验证算法的有效性.结果表明:该方法在信噪比为0 dB时对10种通信调制信号的识别准确率为83.50％,在信噪比为5 dB时对10种通信调制信号的识别准确率为95.25％,该方法识别准确率高、鲁棒性强,可有效应用于无线通信、雷达信号等领域中的信号识别.
Abstract
To solve the problem of unclear features and poor robustness when time-domain features are directly used to identify signals,a communication modulation signal recognition method based on feature distribution density map is proposed.By sampling and analyzing the communication modulation signals,the average value and skewness of the signals are extracted,and the Gaussian kernel function is used to calculate the point probability density to generate the feature distribution density map.Then,deep learning is used for classification and recognition to verify the effectiveness of the algorithm.The results show that the method achieves an identification accuracy of 83.50％for 10 kinds of communication modulation signals at a signal-to-noise ratio of 0 db,and an identification accuracy of 95.25％at a signal-to-noise ratio of 5 db.The proposed method can achieve high identification accuracy and strong robustness.It can be effectively applied to signal identification problems in wireless communication,radar signals,and other fields.
关键词
通信信号/调制识别/特征分布密度图/深度学习算法Key words
communication signal/modulation recognition/feature distribution density map/deep learning algorithm引用本文复制引用
基金项目
国家自然科学基金资助项目(62076160)
上海市自然科学基金资助项目(21ZR1424700)
出版年
2024
NETL
NSTL
万方数据