基于深度残差网络的实时无线电信号监测研究

针对无线电信号监测中传统信号识别与监测方法精度低、可拓展性差、依赖专家特征、难于应用到真实环境中等问题，设计了一种利用深度残差网络结合软件无线电技术进行信号监测的系统。软件无线电技术的可重配置特点使研究人员可以在同一平台部署不同的应用；深度残差网络在调制识别领域的应用展现了出色的性能。但是，很少有研究人员将训练好的调制识别模型直接部署到真实环境中。创新地将深度残差网络与硬件平台结合起来，利用GNU Radio的自定义模块功能，将残差网络嵌入到软件无线电设备中，克服了传统信号部署困难的缺点，证实了残差网络与软件无线电结合的可行性，设计了一种实时信号监测系统，在认知电子战、通信对抗、非协作通信等领域具有重要的应用价值。实验表明，该系统具有实时性强、识别准确率高、易拓展的特点。

Real-Time Radio Signal Monitoring Based on Deep Residual Network

In order to solve the problems of low precision,poor expansibility,dependence on expert features,difficult to apply to real environment in radio signal monitoring,a signal monitoring system based on deep residual network and software radio technology is designed.The reconfigurable feature of software radio technology enables researchers to deploy different applications on the same platform,the application of deep residual network in modulation recognition field shows excellent performance.However,few researchers deploy the trained modulation recognition model directly to the real environment.The deep residual network is innovatively combined with the hardware platform,and the residual network is embeded into the software radio equipment by using the user-defined module function of GNU Radio,which overcomes the shortcomings of traditional signal deployment difficulties,and proves the feasibility of combining residual network with software radio,and a real-time signal monitoring system is designed.The system has important application value in cognitive electronic warfare,communication countermeasures,non-cooperative communication and other fields.Experiments show that the system has the characteristics of strong real-time,high recognition accuracy and easy expansion.