基于混合统计参数识别多种数字信号调制方式  

An Automatic Digital Modulation Recognition Based on Combined Statistic Parameters

作　　者：李纪[1] 何晨[1] 陈杰[1] 

机构地区：[1]上海交通大学电子工程系,上海200240

出　　处：《上海交通大学学报》2007年第5期693-697,共5页Journal of Shanghai Jiaotong University

基　　金：国家自然科学基金资助项目(60372076);上海市科委资助项目(05DZ15004)

摘　　要：为识别QAM等阶数高于4的数字调制信号及提高调制识别算法对高斯白噪声（AWGN）的鲁棒性,在决策论识别参数的基础上,增加了高阶统计量识别参数,并利用混合参数的经向基函数（RBF）神经网络实现数字信号调制样式自动识别,提高了对数字调制信号的识别能力.仿真结果表明,所提出方法对信噪比SNR=-5-20 dB的2ASK、4ASK、8ASK、2PSK、4PSK、8PSK、2FSK、4FSK、8FSK、16QAM和64QAM 11种数字调制测试信号均可以较好地识别.In order to recognize digitally modulated signals with order higher than 4, such as QAM, and to improve the robustness of the recognition method on the AWGN, based on the recognition parameters of the decision-theoretic approach, certain cumulants parameters are added, and the radial basis function （RBF） neural network with those combined parameters is adopted to improve the recognition ability on the digitally modulated signals. With this method, eleven kinds of digital modulation signals, named 2ASK, 4ASK, 8ASK, 2PSK, 4PSK, 8PSK, 2FSK, 4FSK, 8FSK, 16QAM and 64QAM, can be recognized with approving results when SNR is clown to -5 dB.

关 键 词：信号识别 混合统计参数 高阶累积量 径向基函数神经网络 

分 类 号：TN911.7[电子电信—通信与信息系统]