Radar emitter signal recognition based on stackedauto-encoder and ambiguity function main ridge

Aiming at the limitation of artificial extraction of radar emitter signal features, such as long extraction period and incomplete feature description, a radar emitter signal recognition method is proposed based on ambiguity function main ridge and stacked auto-encoder of deep learning. According to the ambiguity function main ridge contains rich intrinsic modulation information; the method of this paper extracts abstract features for classification recognition from radar emitter signals. We experiment with six radar emitter signals and comparesthe proposed method with other artificial features and deep learning methods. The experimental results show that the proposed method can maintain 100% recognition accuracy when the signal-noise ratio is above 2 dB. And the signal-noise ratio is -6 dB, the recognition accuracy can still be maintained above 82.83%, which is significantly higher than other methods. Eventhough in the signal environment that contains the same modulation type but different modulation parameters, when the signal-to- noise ratio is greater than 0 dB, the recognition accuracy is stable above 95%. The recognition accuracy can reach 79% when the signal-noise ratio is reduced to -4 dB.It is proved that the method can effectively extract the deep features of the signal and has good anti-noise performance, which basically meets the requirements of the actual battlefield.