基于小波变换和神经网络模型的数字调制识别方法

SCHEME OF DIGITAL MODULATION RECOGNITION BASED ON WAVELET TRANSFORM AND NEURAL NETWORKS

由于小波变换对瞬态信息具有较强的检测能力,数字调制信号在间断点呈现不同的瞬态信息。使用提取小波变换后包络方差与均值平方之比的特征参数,来实现3种信号(MFSK、MPSK和MQAM)的类间识别。然后提取经小波变换后的信号幅度层数N1,对MFSK进行类内识别,提取经归一化后的信号再经过小波变换后的尖峰数N2,对MPSK进行类内识别。最后利用人工神经网络作为分类器,仿真结果表明在低信噪比下具有良好的正确识别率。

Due to the strong detection capability of wavelet transform （WT） on transient information, digitally modulated signals present different transient information on discontinuity points. First, the characteristic parameter of the ratio of envelope variance to mean square after WT is extracted to realise inter-category recognition of 3 signals （ MFSK, MPSK and MQAM）. Then, through the extraction of signal amplitude layer N1, the intra-class recognition of MFSK is achieved. Thirdly, the peak number N2, which is attained from the normalised signals undergoing WT again, is extracted for intra-class recognition of MPSK. Lastly, the artificial neural network is employed as the classifier. Simulation results demonstrate that this scheme has good accurate recognition rate in the condition of low signal-to-noise ratio.