A method and device for identifying a signal modulation mode

Abstract

The present invention provides a signal modulation identification method. First, the original signal is mapped from the time domain to the frequency domain by wavelet transform, and then the wavelet coefficient map of the signal is constructed, and then the signal wavelet is extracted by using the VGG convolutional neural network combined with the self-encoder. Coefficient image features, calculate the CORAL loss between the known sample signal feature set and the unknown sample signal feature set, calculate the classification loss of the known sample signal set, and jointly optimize the sum of the two losses to make the VGG convolutional neural network parameters reach Optimum, so that the neural network can identify the unlabeled target domain signal; the present invention also provides a signal modulation identification device, including a signal receiving module, a digital module, a data processing module and a power supply; the method of the present invention can receive non- The unlabeled modulated signal of the partner is trained in conjunction with the labeled signal of the own party to achieve the recognition of the unlabeled non-partner signal.

Description

technical field [0001] The invention relates to the technical field of signal identification, in particular to a signal modulation mode identification method and a signal modulation mode identification device. Background technique [0002] Modulation identification, also known as modulation classification, refers to the classification of the modulation type of the received signal. Communication modulation identification technology is of great significance to improve the efficiency of communication countermeasures. Modulation identification technology can assist in determining the type of enemy communication sources, and then can assess the threat level of enemy radio stations. The traditional communication modulation recognition technology is based on the time-frequency distribution, instantaneous amplitude, phase, frequency, high-order statistics, and cyclic spectrum of the communication signal as the classification characteristic parameters. [0003] Different from the e...

AI Technical Summary

Problems solved by technology

[0004] Although the modulation classification technology based on the above-mentioned convolutional neural network has a high recognition rate and robustness, it requires a large amount of labeled training data as support.

In the actual use environment, there will always be new, unlabeled target domain signals that are distributed differently from the training data at the signal receiving end. At this time, it is difficult to accurately identify the pre-trained neural network using the labeled source domain signal This new class and unlabeled signal

Images