基于深度卷积神经网络的数字调制方式识别

Digital modulation recognition based on deep convolutional neural network

摘要:

针对非协作通信条件下信号调制方式识别问题,提出了一种基于深度神经网络的调制方式自动识别新方法。该方法对接收到的信号进行预处理,生成星座图,并将星座图形状作为深度卷积神经网络的输入,根据训练好的网络模型对调制信号进行分类识别。与以往的识别方法相比,该方法利用卷积神经网络自动学习各种数字调制信号的星座图特征,克服了特征提取困难,通用性不强,抗噪声性能差等缺点,处理流程简单,并对星座图的形变具有不敏感性。针对4QAM、16QAM和64QAM三种典型的数字调制方式,进行了仿真实验,当信噪比大于4时,调制方式的识别正确率大于95％,实验结果表明,基于深度卷积神经网络的信号调制方式识别方法是有效的。

Abstract:

A novel method of automatic modulation recognition in non-cooperation communication systems, which is based on deep convolutional neural network, is proposed. Firstly, the received signal is preprocessed and generates the constellation diagram. Then, the shape of the constellation diagram is used as the input of the deep convolution neural network, which is trained to classify the modulated signal. The convolution neural network can automatically learn the constellation diagram features of various digital modulation signals, which can simplify the processing procedures and overcome the weaknesses of traditional techniques, such as the difficulty in extracting the features, the absence of universal property, and the poor anti-noise performance. In addition, the deformation of the constellation diagram is insensitive to the final classification performance by using convolution neural network. Three typical digital modulation schemes including 4QAM, 16QAM and 64QAM are used in the simulation test, and the results show that when the SNR is greater than 4, the accuracy of modulation recognition is more than 95%, which confirmed that the proposed method is effective.