An Efficient Automatic Modulation Classifier based on VGG Net Model

Automatic modulation classification is gaining significant importance in the field of communication intelligence gathering. In this paper, we present a deep learning model based on visual geometry group network for automatic modulation classification. Two widely used benchmark datasets-RadioML2016.10a and RadioML2016.10b are used for training and testing of the proposed classifier model. Through comprehensive simulation results, it is shown that the proposed model achieves a significant performance improvement in classification accuracy. It offers the modulation classification accuracy of 97.4% at high signal-to-noise-ratio and 95.2% at low signal-to-noise-ratio ranging over 6 dB to 0 dB respectively. It is also observed that the proposed classifier yields superior performance at RadioML2016.10b as compared to the former version of the dataset.