Modulation Recognition Based on IQ-eyes Diagrams and Deep Learning

This article presents our initial results in deep learning for modulation recognition with the signal IQ-eyes diagrams. we design a representative convolution neural network to distinguish between eight modulation classes which shows more promising when compared to more traditional likelihood-based or feature-based techniques. Furthermore, we discuss the impact of frequency offset and influence of network parameters on recognition performance though the recognition accuracy. Moreover, the method based on multi-terminal processing can mine the intrinsic characteristics of the signal more completely, which reduces the requirement of network and makes the network more robust.