Multidimensional CNN-LSTM Network for Automatic Modulation Classification

Automatic modulation classification (AMC) is the premise for signal detection and demodulation applications, especially in non-cooperative communication scenarios. It has been a popular topic decades gained significant progress with development of deep learning methods. To further improve accuracy, hierarchical multifeature fusion (HMF) based on multidimensional convolutional neural network (CNN)-long short-term memory (LSTM) proposed this paper. First, CNN module (MD-CNN) feature compensation between interactive features extracted by two-dimensional filters respective one-dimensional filters. Second, learnt MD-CNN are fed into an LSTM layer exploitation temporal features. Finally, results obtained Softmax classifier. The effectiveness method verified abundant experimental two public datasets, RadioML.2016.10a RadioML.2016.10b. Satisfying as compared state-of-the-art