Toward receiver, modulation, carrier and symbol rate agnostic SEI Dataset

Specific emitter identification (SEI)  is a promising authentication paradigm in physical layer security (PLS). Despite the significant success of existing SEI schemes, most of them assume that the distributions of the training dataset and the test dataset are consistent. However, in most practical scenarios, when the signal parameters change, the distribution of the samples will changes,  resulting in a significant performance degradation. To this end, we construct a SEI dataset with variable parameters, including variable receiver, variable modulation, variable carrier and variable symbol rate. Compared to existing public works, the constructed  dataset comprehensively considers the impact of different signal parameters on SEI tasks, providing data support for subsequent research work. 

特定辐射源识别（Specific Emitter Identification，SEI）是物理层安全（Physical Layer Security，PLS）中极具前景的认证范式。尽管现有的特定辐射源识别方案已取得显著进展，但其中大多数均假设训练数据集与测试数据集的分布保持一致。然而在多数实际场景中，当信号参数发生变化时，样本分布亦会随之改变，进而导致模型性能出现显著退化。为此，我们构建了一款参数可变的特定辐射源识别数据集，涵盖可变接收机、可变调制方式、可变载波以及可变符号率四类参数变化场景。相较于现有公开研究成果，本数据集全面考量了不同信号参数对特定辐射源识别任务的影响，可为后续相关研究工作提供数据支撑。