Recently, the latest findings of Academician Su Donglin’s team from the School of Electronics and Information Engineering at Beihang University, titled “Single sample electromagnetic spectrum recognition utilizing fractional Fourier transform,” have been published in Chinese Journal of Aeronautics, an open access, peer-reviewed international journal in the fields of aeronautics and astronautics.

Lu Xiaozhu, a PhD candidate from the School of Electronics and Information Engineering, is the first author of the paper, while Associate Professor Song Lingnan serves as the corresponding author.

Electromagnetic Spectrum (EMS) recognition is vital in spectrum control, interference location, electronic countermeasures, etc. However, samples of high-value targets are incredibly scarce, even single, and are easily overwhelmed by noise and numerous low-value targets, resulting in poor recognition accuracy using traditional methods. Furthermore, the great similarity between samples from the same manufacturer, model, and batch, makes Specific Emitter Identification (SEI) with the EMS especially challenging.

In the study, based on the powerful extension and extraction ability of the Fractional Fourier Transform (FrFT) for detailed features, the researchers propose a novel algorithm for the EMS recognition under a single-sample condition (see Fig.1). The proposed method constructs a feature matrix FrFT-M from the results of the FrFT under specific orders for each sample. Then, the most relevant item, obtained by analyzing the correlations among FrFT-Ms between the unidentified sample and known samples, determines the optimal recognition. Three simple tests are conducted, including two simulations considering fifteen basic waveforms and six typical radar signals, and one experiment using STM32 microcontroller boards. The detection results of simulated and experimental data show that the accuracies of all three cases are higher than 86%, even for samples of the same model. The method is promising and may have significant value in other fields.

Fig.1 Framework of the proposed single-sample recognition method

This work was supported in part by the National Natural Science Foundation of China.

Link to the article: https://doi.org/10.1016/j.cja.2024.01.024

Editor: Lyu Xingyun