Multicomponent Radar Signal Recognition via Signal Quantity Detection and Feature Enhancement

Radar signal recognition under low Signal-to-Noise Ratio (SNR) constitutes a crucial aspect of modern electronic reconnaissance systems. With the increasing complexity of the electromagnetic environment, accurately recognizing mixed radar signals with multiple components has become a formidable challenge. To address this issue, we propose a novel radar signal recognition method based on signal quantity detection and feature enhancement. This method first employs a signal quantity detection module to estimate the number of signals present in the received sample and then routes the sample to corresponding recognition models based on the determined signal quantity. The signal recognition model is composed of two key components: the feature enhancement module and the reverse difference module. The feature enhancement module is designed to improve signal quality and optimize feature extraction. The reverse difference module further amplifies the distinctions between different signals through reverse difference training, thereby enhancing overall recognition performance. Experimental evaluations on a mixed signal dataset with varying numbers of signals demonstrate the effectiveness of the proposed method. At an SNR of -6dB, our approach achieves over 95% recognition accuracy, showcasing its substantial potential and practical value.