| 摘要: |
| 非合作接收条件下的通信信号检测与调制识别是展开无线电频谱监测和战场通信侦察对抗的重要基础。在当前的无线电监测任务中不仅需要识别特定频段内信号的存在，还涉及到对信号调制类型的识别。然而，传统的信号检测与调制识别往往是作为两个独立的研究领域进行探讨的，缺乏将两者联合起来研究的方法。提出了一种基于时间卷积长短时记忆神经网络的深度学习框架，将信号检测与调制识别联合研究。该框架突破传统分立研究范式，通过时间卷积网络（Temporal Convolutional Network，TCN)与长短期记忆网络（Long Short-term Memory，LSTM）的级联结构实现信号检测与调制识别功能融合：利用TCN捕捉信号时序特征，结合LSTM的长期依赖建模能力，构建二分类信号检测模块与多分类调制识别模块的协同工作机制。在基准数据集RML2016.10a下的信号检测和调制识别实验表明，信号检测平均概率提高了6%~13%，调制识别平均准确率提高了0.16%~7.53%。 |
| 关键词: 通信信号检测 调制识别 时间卷积网络 长短时记忆网络 |
| DOI：10.20079/j.issn.1001-893x.241119001 |
|
| 基金项目:国家自然科学基金资助项目（61801319）；四川省自然科学基金面上项目（2026NSFSC0394）；数字化学习技术集成与应用教育部工程研究中心项目(1321002)；四川轻化工大学2022研究生创新基金(Y2023288) |
|
| Wireless Communication Signal Detection and Modulation Recognition Algorithm Based on TLNet |
| ZHANG Xueqin琣,LUO Zhongqiang,LAN Mengxian |
| (a.School of Automation and Information Engineering;b.Intelligent Perception and Control Key Laboratory of Sichuan Province，Sichuan University of Science and Engineering，Yibin 644000，China) |
| Abstract: |
| Communication signal detection and modulation recognition under non-cooperative reception conditions are an important basis for radio spectrum monitoring and battlefield communication reconnaissance and confrontation.In current radio monitoring tasks，it is necessary to identify not only the presence of a signal in a specific frequency band，but also the type of signal modulation.However，traditional signal detection and modulation recognition are often discussed as two independent research fields，and there is a lack of methods to combine the two.In this study，a deep learning framework based on temporal convolutional long short-term memory(LSTM) neural network is proposed，which combines signal detection and modulation recognition.This framework breaks through the traditional discrete research paradigm and realizes the integration of signal detection and modulation recognition functions through the cascade structure of temporal convolutional network(TCN) and LSTM.The collaborative working mechanism of the binary classification signal detection module and the multi-classification modulation and identification module is constructed.The signal detection and modulation recognition experiments under the benchmark dataset RML2016.10a show that the average probability of signal detection is increased by 6% |
| Key words: communication signal detection modulation recognition time convolutional network long short-term memory network |
《电讯技术》编辑部 