Abstract:

The modulation recognition of communication signals is an important research issue in non-cooperative fields. A method based on entropy features and support vector machine (SVM) for modulation recognition is put forward. Multidimensional entropy features are extracted as the input coefficients of the classifier. The performance of the binary tree based SVM classifier is investigated using these entropy features. Besides signal-to-noise ratio, the algorithm needs no further information for the received signal such as signal bandwidth or carrier frequency. Theoretical analysis and simulation results show that the algorithm is practically valuable for its high recognition accuracy and less computation load.