基于卷积神经网络的3“A”信号分类

3A signal classification based on convolutional neural network

信号分类是一种重要的无线电技术,相比传统的分类方法,基于深度学习的方法具有更高的准确率和鲁棒性,可以应用于更加复杂的信号环境中。为了更好的区分ACARS、AIS、ADS-B这3种海空通信信号,本文提出了一种基于卷积神经网络的3“A”信号分类方法。将3种信号的I/Q数据加入噪声,转变为二维信号输入卷积神经网络实现对信号的分类。实验结果表明,在测试集上得到90%的准确率,对比传统基于特征的分类方法,该网络对信号的分类效果较好。

Signal classification is an important wireless technology.Compared to traditional classification methods,deep learningbased methods have higher accuracy and robustness,and can be applied to more complex signal environments.In order to better distinguish these three sea-air communication signals,namely ACARS,AIS,and ADS-B,this article proposes a 3"A"signal classification method based on convolutional neural network(CNN).The I/Q data of the three signals are added with noise and transformed into two-dimensional signals for input to the CNN for signal classification.Experimental results show that the network achieved an accuracy of 90%on the test set,and compared with traditional feature-based classification methods,the network had better performance in signal classification.