Modulation Mode Recognition of Communication Signal Based on FACAM-Efficient Net
In view of the problems such as noise interference,insufficient recognition rate,and complex network model in communication signal recognition methods in complex electromagnetic space,the FACAM-Efficient Net signal modulation recognition method based on high-order cumulant visualization time-frequency images is proposed.Through high-order cumulant time-frequency analysis method,the communication signals of different modulations are converted into visualization time-frequency images for the training and testing of network models.According to the image features of visualization time-frequency images,the FACAM frequency-domain channel hybrid attention mechanism is proposed,and FACAM is introduced to improve the network main structure of Efficient Net so as to better extract the features of frequency-domain and channel of time-frequency images.Experimental results show that the modulation recognition method proposed in this paper has a good recognition effect at low signal-to-noise ratio.
high-order cumulanttime-frequency analysismodulation recognitionimproved Efficient Net
NETL
NSTL
万方数据
