Review on the Development of Few-Shot Specific Emitter Identification Technology
Dian Lv, Zhiyong Yu, Jiawei Xie, Hao Zhang · 2024
In the face of a complex and ever-changing electromagnetic environment, the rapid increase in the number and types of radiation sources has made traditional Specific Emitter Identification techniques, which rely on conventional machine learning and statistical features, difficult to apply. In recent years, deep learning models have demonstrated exceptional performance in areas such as image recognition, prompting many researchers to incorporate deep learning into radiation source identification tasks. However, deep learning requires vast amounts of high-quality data, and in practical electromagnetic environments, it is challenging to collect and annotate radiation source signals. As a result, few-shot specific emitter identificatio has become a new focal point of research. This paper first introduces the radiation source identification system and reviews the research on traditional radiation source identification techniques. Next, it provides a detailed overview of few-shot specific emitter identification techniques categorized based on the radiation source identification technology framework, including data augmentation, metric learning, and model-based methods. Finally, the paper analyzes the current challenges in Few-Shot Specific Emitter Identification technology and offers prospects for future research.