Digital Modulation Recognition Method Based on Tree-Structured Neural Networks

摘要：

This paper is focusing on the neural network based classifier design of modulation types for communication signals. A tree-structured neural network is proposed which could make correct identification among 13 modulation types by the use of comprehensive features, including power spectral features, cyclic spectral features and high-order cumulant features. The tree-structured neural network is a self-organizing, hierarchical classifier implementing a sequential linear strategy and requiring no statistical analysis of the features. The design procedure is discussed and simulation results are presented. Experiments show that these types of modulation can be recognized under low SNr in AWGN, and this method also works well for frequency modulations and some amplitude-phase modulation in multipath environment.

关键词： neural network modulation recognition power spectral feature cyclic spectral feature cumulant feature

作者: Yiqiong Xu Lindong Ge Bo Wang

作者单位: National Digital Switching System Engineering and Technology Research Center Zhenzhou, China National Digital Switching System Engineering and Technology Research Center Zhenzhou,China

会议类型: 国际会议

会议名称: The International Conference on Communication Software and Networks(2009 IEEE通信软件与网络国际会议 ICCSN 2009)

会议地点: 成都

会议语种:英文

页码: 708-712

在线出版日期: 2009-02-20（万方平台首次上网日期，不代表论文的发表时间）