基于深度学习的雷达辐射源识别算法

为解决雷达辐射源识别中特征提取困难、低信噪比条件下识别效率低的问题,提出了一种基于一维卷积神经网络和长短期记忆网络的深度学习智能识别算法,构建了一个CNN?LSTM网络,能实现对不同脉内调制方式雷达辐射源的端到端识别。该网络首先利用卷积层学习信号局部特征,然后将卷积层输出的结果输入长短期记忆网络,学习信号的全局特征,最终构造逻辑回归分类完成分类识别任务。仿真结果表明,该算法较单一卷积神经网络模型具有更好的识别效果,抗噪声效果更强,在-6 dB信噪比的条件下,识别的准确率仍能够达到90%以上。

Radar emitter identification algorithm based on deep learning

In order to solve problems of feature extraction and low recognition efficiency under low signal-tonoise ratio in radar emitter identification,a deep learning intelligent recognition algorithm is proposed based on one-dimensional convolutional neural network and long short-term memory(CNN-LSTM)network.A CNNLSTM network is constructed,which can realize end-to-end identification of different pulse modulation radar sources.The convolutional layers are used to learn the local characteristics of the signal by this network,and the long short-term memory layer is used to learn the global characteristics.Finally a logistic regression classification is constructed to complete the recognition task.Simulation results show that the algorithm has better recognition effect and stronger anti-noise effect compared with the single convolutional neural network model.Under the condition of-6 dB signal-to-noise ratio,the recognition accuracy can still reach more than 90%.