Recognition of modulation signal based on a new method of feature selection
新特征选择方法下的信号调制识别

According to the characteristic of communication signals, a new method of feature selection is presented. In this method, the efficient and robust genetic algorithm is used to sift out qualified feature subset, under the rule of clustering level of same classifications and separating level of different classifications. With the fine structure of evidence theory, a considerable recognition ratio is obtained under a wide range of SNR environment. Finally, the effectiveness of this method is proved by computer simulation.