Automatic Modulation Recognition Using Modified Convolutional Neural Network

With the development of Deep Learning as a potent method for separating high-level representations from complex data, numerous fields have seen substantial progress. A crucial task in cognitive radio systems, spectrum sensing, and signal intelligence is automatic modulation recognition. The modulation type of a certain radio broadcast can be determined with the help of deep learning methods. In this study, we have used the RadioML 2016.10b dataset to evaluate the performance of convolutional neural networks (CNNs) for automatic radio modulation recognition. The CNN architecture is tuned to optimize the performance. The combination of six convolutional layers and two dense layers gives an accuracy of approximately 93.334% at 12dB.