A IQ Correlation and Long-term Features Neural Network Classifier for Automatic Modulation Classification

Automatic modulation classification is the process of determining the modulation scheme of an intercepted signal with no a prior information. Its main advantage is that no communication overhead needs to be allocated for control information to inform the receiver about changes in a transmitted signal's modulation scheme. Recently, deep learning has been applied to automatic modulation classification tasks, but most deep learning classifiers have the difficulty balancing weight and efficiency. According to the characteristics of in-phase (I) and quadrature (Q) signals, a IQ correlation and long-term temporal features neural network classifier is proposed. It first extracts IQ correlation features and short-term temporal features by different directional convolutional filters, and then extracts long-term temporal features by the Long Short-Term Memory unit, and finally uses only one full connected layer for classification. Experiment results in the standard dataset RML2016.10a show that our proposed is more efficient and lightweight compared with the state-of-the-art deep learning classifiers.