| 摘要: |
| 针对现有通信干扰信号识别方法识别效果不佳的问题，提出了一种基于长短时记忆网络（Long ShortTerm Memory，LSTM）和特征融合的通信干扰识别方法。该方法利用LSTM网络提取干扰信号的特征，通过LSTM强大的序列特征提取能力提升干扰信号特征提取的性能；通过提取信号的时域和频域特征后进行特征融合，使用全连接分类器对干扰信号进行分类识别，提升特征提取的完整性和干扰识别的性能。仿真表明，所提方法的干扰识别性能相比于现有的基于卷积神经网络的干扰识别方法提升了6 dB，可用于通信干扰信号类型的识别。 |
| 关键词: 干扰识别 长短时记忆（LSTM）网络 特征融合 深度学习 |
| DOI： |
|
| 基金项目: |
|
| Communication jamming signals recognition based on LSTM network and feature fusion |
| WEI Di,ZENG Haibin,HONG Feng,MA Song,YUAN Tian |
| (1.Southwest China Institute of Electronic Technology,Chengdu 610036,China;2.Beijing Institute of Tracking and Telecommunication Technology,Beijing 100094,China;3.Unit 63750 of the PLA,Xian 710043,China;4.National Key Laboratory of Science and Technology on Communications,University of Electronic Science and Technology of China,Chengdu 611731,China) |
| Abstract: |
| For the poor recognition effect of existing communication jamming signals recognition methods,a recognition method based on long short-term memory(LSTM) network and feature fusion is proposed.In this method,the LSTM network is used to extract signal features.With the powerful sequence feature extraction capability of the LSTM network,this method can improve the feature extraction ability of jamming signals.The time-domain and frequency-domain features of the signals are extracted through the feature extraction network,and are fused through the fusion module,and the fusion feature is used to identify the signals through the fully connected classifier,which can improve the integrity of feature extraction and the performance of jamming signals recognition.The simulation results demonstrate that the jamming signals recognition performance of the proposed method is improved by 6 dB compared with the existing jamming signals recognition method based on convolutional neural network,and it can be applied in the recognition of communication jamming signals types. |
| Key words: jamming signals recognition long short-term memory(LSTM) network feature fusion deep learning |
