Modulation Recognition of Composite Signal Based on ResNet and Frequency Domain Graph

Abstract

Aiming at the problem of composite signal recognition in measurement and control system, a modulation recognition method of composite signal is proposed based on clipped internal modulation signal frequency image and residual network. Residual network has the advantages of fast convergence and high recognition accuracy, which is used to realize the recognition of different modulation modes and different bandwidth signals. The experimental results show that, compared with the traditional decision tree algorithm and the constellation based CNN recognition algorithm, the recognition performance of the proposed method is improved by more than 3dB for the variable bandwidth composite modulation signal set (BPSK-FM and QPSK-FM with internal modulation bandwidth of 128kHz, BPSK-FM and QPSK-FM with internal modulation bandwidth of 2kHz and 2FSK-FM with internal modulation bandwidth of 32kHz and 2kHz).