Research on Radiation Source Identification Technology Based on Deep Learning

With the rapid development of artificial intelligence, deep learning has also been introduced into the field of communication for the study of signal modulation recognition, radar signal recognition, signal emitter identification, radar interference identification, etc. In this paper, the spectral probability statistics technology is used to obtain the spectral characteristics of the emitter, and the YOLOv5 target detection algorithm is optimized, and more recognizable features are fused by expanding the receptive field, which strengthens the sensitivity to small changes in the spectral spectrum of the emitter, improves the recognition accuracy of the spectral texture boundary of the emitter, and ensures that the model is lightweight and can meet the needs of emitter identification in complex interference environments.