Deep Learning Based Automatic Modulation Classification in the Case of Carrier Phase Shift

Various tools and methods have been used in the problem of automatic modulation classification (AMC) with success, such as maximum likelihood estimation (MLE), K-nearest neighbor (KNN), genetic programming or deep learning. In this paper, we consider another perspective: efficiency. We proposed a novel polar coordinate approach in deep learning to mitigate the adverse effects of carrier phase shift without using it in the training of deep learning networks. Thus, we can reduce the amount of training data and the amount of training time.