A Deep Learning approach for Modulation Recognition

Abstract

Communication signal modulation recognition refers to a process of automatically processing a received signal and determining its modulation type. As an intermediate part of signal detection and demodulation, modulation recognition technology plays an important role in a variety of civilian and military applications such as cognitive radio, electronic reconnaissance, and intelligent demodulator. After several decades of development, modulation recognition technology has made many achievements. However, with the increasing demand for engineering and the increasingly complex wireless communication channel environment, there are still many issues to be solved. This paper presents a method of deep learning for modulation recognition. Deep learning allows for automatically learn features directly from simple wireless signal representations, without the need to design of hand-crafted expert features such as high-order cyclic moments. The neural network used in this paper is the deep belief network. Experimental result shows that using the temporal IQ data representation to recognize modulation formats contributes to the high correct recognition rate at high SNR. When the SNR is $18\mathrm {d}\mathrm {B}$, the average recognition rate still reaches 92.12%.