软件无线电中数字调制信号的自动识别

Automatic identification of digital modulation for software radio

针对软件无线电系统中数字信号调制方式的自动识别提出一种改进算法.使用从调制信号中提取的的特征参数,对多个MLP分类器进行独立训练,使用输出向量的加权投票表决法来融合多个MLP分类器的方法,得到了在信噪比大于等于10dB时良好的效果,验证了方法的可行性.

A novel algorithm for automatic identifiction of digital modulation is proposed,which can be used in the field of software radio. The mehod is: some high quality features are extracted from the modulated signals, a number of MLP classifiers are trained seperately and independently, a new weighted voting mothod,the weight value of which is based on the output vectors of the classifiers,is adopted to accomplish the fusion of multiple classifier. Simulation results demonstrate that this method is efficient when SNR≥10 dB.