基于CLDNN的调制信号识别方法  

MODULATION SIGNAL RECOGNITION METHOD BASED ON CLDNN

作　　者：张军[1] 符杰林[1] 林基明 Zhang Jun;Fu Jielin;Lin Jiming(Key Laboratory of Cognitive Radio&Information Processing of the Ministry of Education,Guilin University of Electronic Technology,Guilin 541004,Guangxi,China;Guangxi Colleges and Universities Key Laboratory of Satellite Navigation and Position Sensing,Guilin 541004,Guangxi,China)

机构地区：[1]桂林电子科技大学认知无线电与信息处理教育部重点实验室,广西桂林541004 [2]广西高校卫星导航与位置感知重点实验室,广西桂林541004

出　　处：《计算机应用与软件》2021年第10期216-220,277,共6页Computer Applications and Software

摘　　要：调制信号的识别在军用的电子战和民用的智能化无线通信中占有重要的地位,针对现有识别方法识别种类少、整体识别率不高和需要预处理等缺点,设计一个CLDNN端到端深度神经网络。该网络无需人工干预或数据统计,自动提取特征并进行多类调制信号类型识别。实验结果表明,该方法能够同时识别11种信号的调制方式,在低信噪比下识别效率相比现有方法有所提升,当信噪比在-4 dB以上时,整体识别精度达到94%以上。The modulation signal recognition plays an important role in military electronic warfare and civilian intelligent wireless communication.Aiming at the shortcomings of the existing recognition methods,such as low number of recognition types,low overall recognition rate,and the need for preprocessing,a CLDNN end-to-end deep neural network is designed.The network could extracte features and recognize multiple types of modulated signals automatically without manual intervention or data statistics.The experimental results show that this method can identify 11 signal modulation modes at the same time,and the recognition efficiency is improved compared with the existing methods under low signal-to-noise ratio.When the signal-to-noise ratio is above-4 dB,the overall recognition accuracy reaches over 94%.

关 键 词：信号识别 端到端 神经网络 CLDNN 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]