Embedded RF fingerprint interpretation: multi-channel complex residual networks with adaptive sphere space decision boundaries

doi:10.23919/JSEE.2025.000155

Journal of Systems Engineering and Electronics ›› 2026, Vol. 37 ›› Issue (1): 137-147.doi: 10.23919/JSEE.2025.000155

• DEFENCE ELECTRONICS TECHNOLOGY • Previous Articles Next Articles

Embedded RF fingerprint interpretation: multi-channel complex residual networks with adaptive sphere space decision boundaries

Yongsheng DUAN(), Junning ZHANG(), Lei XUE(), Ying XU()

College of Electronic Engineering, National University of Defence Technology, Hefei 230037, China

Received:2025-05-15 Accepted:2025-09-11 Online:2026-02-18 Published:2026-03-09
Contact: Junning ZHANG E-mail:406810103@qq.com;zjn20101796@sina.cn;eeixuelei@163.com;eeixuying@163.com
About author:
DUAN Yongsheng was born in 1991. He received his M.S. degree in pattern recognition and intelligent systems from Naval Command College, Nanjing, China, in 2014. He is working toward his Ph.D. degree in the College of Electronic Engineering, National University of Defense Technology, Hefei, China. His research interests mainly include target detection, tracking, and recognition. E-mail: 406810103@qq.com

ZHANG Junning was born in 1992. He received his Ph.D. degree in Army Engineering University of PLA, Shijiazhuang, China, in 2020. He is an associate professor in the College of Electronic Engineering, National University of Defense Technology, Hefei, China. His current research interests include target detection, tracking, and recognition. E-mail: zjn20101796@sina.cn

XUE Lei was born in 1963. He received his B.S. and M.S. degrees in electrical engineering from Electronic Engineering Institute, Hefei, China, in 1983 and 1990, respectively. He is a professor in the College of Electronic Engineering, National University of Defense Technology, Hefei, China. His current research interests include signal processing and its applications. E-mail: eeixuelei@163.com

XU Ying was born in 1979. She received her Ph.D. degree in electrical engineering from Electronic Engineering Institute, Hefei, China. She is an associate professor in the College of Electronic Engineering, National University of Defense Technology, Hefei, China. Her current research interests include information fusion and situation analysis. E-mail: eeixuying@163.com
Supported by:
This work was supported by the National Natural Science Foundation of China (62201602).

Abstract

Abstract:

Despite the superior advantages of specific emitter identification in extracting emitter features from in-phase and quadrature (I/Q) signals, challenges persist due to signal-type confusion and background noise interference. To address those limitations, this paper proposes a multi-channel contrast prediction coding and complex-valued residuals network (MCPC-MCVResNet) framework. This model employs contrast prediction techniques to directly extract discriminative features from electromagnetic signal sequences, effectively capturing both amplitude and phase information within I/Q data. A core innovation of this approach is the sphere space softmax (SS-softmax) loss, which optimizes intra-class clustering density of while establishing well-defined boundaries between distinct emitters. The SS-softmax mechanism significantly enhances the model’s capacity to discern subtle variations among radiation emitters. Experimental results demonstrate superior identification accuracy, rapid convergence, and exceptional robustness in low signal-to-noise ratio environments.

Key words: specific emitter identification (SEI), multi-channel complex-valued residual network (MCVResNet), sphere space softmax (SS-softmax)

Yongsheng DUAN, Junning ZHANG, Lei XUE, Ying XU. Embedded RF fingerprint interpretation: multi-channel complex residual networks with adaptive sphere space decision boundaries[J]. Journal of Systems Engineering and Electronics, 2026, 37(1): 137-147.

Figures/Tables 8

Fig 1

Fig 2

Fig 3

Fig 4

Fig 5

Table 1

Fig 6

Fig 7

References 34

1	ZHANG J N, LIU Y, DING G R, et al Adaptive decomposition and extraction network of individual fingerprint features for specific emitter identification. IEEE Trans. on Information Forensics and Security, 2024, 19, 8515- 8528. doi: 10.1109/TIFS.2024.3427361
2	CHEN J, ZHOU R P, GUI W H A serial game distributed algorithm to ϵ-minimum vertex cover of networks. IEEE Trans. on Automation Science and Engineering, 2025, 22, 1534- 1542. doi: 10.1109/TASE.2024.3367722
3	ZHAO Y R , WANG X , SUN L T, et al Specific emitter identification based on frequency and amplitude of the signal kurtosis. Journal of Systems Engineering and Electronics, 2025, 36 (2): 333- 343. doi: 10.23919/JSEE.2023.000054
4	FAN R , SI C K , HAN Y, et al RFFsNet-SEI: a multidimensional balanced-RFFs deep neural network framework for specific emitter identification. Journal of Systems Engineering and Electronics, 2024, 35 (3): 558- 574. doi: 10.23919/JSEE.2023.000069
5	FANG Y Y, WEI S, ZHAO Y, et al Radar-specific emitter identification with only envelope power based on multidimensional complex noncentral Chi-square classifier. IEEE Sensors Journal, 2023, 23 (17): 20223- 20235. doi: 10.1109/JSEN.2023.3298352
6	CHEN Y C, LI P, YAN E X, et al A knowledge graph-driven CNN for radar emitter identification. Remote Sensing, 2023, 15 (13): 3289. doi: 10.3390/rs15133289
7	ZHENG C, WU H, HAO Y, et al Radar emitter identification based on deep learning of inter-pulse features. Journal of Signal Processing, 2020, 36 (8): 1187- 1195.
8	CHURCHILL F E, OGAR G W, THOMPSOM B J The correction of I and Q errors in a coherent processor. IEEE Trans. on Aerospace and Electronic Systems, 1981, 17 (1): 131- 137.
9	BASSAM S A, BOUMAIZA S, GHANNOUCHI F M Block-wise estimation of and compensation for I/Q imbalance in direct-conversion transmitters. IEEE Trans. on Signal Processing, 2009, 57 (12): 4970- 4973. doi: 10.1109/TSP.2009.2026598
10	WANG Y, GUI G, GACANIN H, et al An efficient specific emitter identification method based on complex-valued neural networks and network compression. IEEE Journal on Selected Areas in Communications, 2021, 39 (8): 2305- 2317. doi: 10.1109/JSAC.2021.3087243
11	SINHA H, AWASTHI V, AJMERA P K Audio classification using braided convolutional neural networks. IET Signal Processing, 2020, 14 (7): 448- 454. doi: 10.1049/iet-spr.2019.0381
12	KOBAL T, LI Y, WANG X, et al Digital predistortion of RF power amplifiers with phase-gated recurrent neural networks. IEEE Trans. on Microwave Theory and Techniques, 2022, 70 (6): 3291- 3299. doi: 10.1109/TMTT.2022.3161024
13	MOSSAD O S, ELNAINAY M, TORKI M. Deep convolutional neural network with multi-task learning scheme for modulations recognition. Proc. of the 15th International Wireless Communications & Mobile Computing Conference, 2019: 1644−1649.
14	HENNEKE L. Improving data-driven RF signal separation with SOI-matched autoencoders. Proc. of the IEEE International Conference on Acoustics, Speech, and Signal Processing Workshops, 2024. DOI: 10.1109/ICASSPW62465.2024.10626245.
15	YADAV H, SITARAM S, SHAH R R. Analysing the masked predictive coding training criterion for pre-training a speech representation model. Proc. of the IEEE International Conference on Acoustics, Speech and Signal Processing, 2023. DOI: 10.1109/ICASSP49357.2023.10096167.
16	WEI P L, TONG M I Attention-DPU: dual-path UNet with an attention mechanism for ultrasound image segmentation. Journal of Physics: Conference Series, 2020, 1693 (1): 012155. doi: 10.1088/1742-6596/1693/1/012155
17	ZHANG J N, WEI Z, DING G, et al Multi-channel lightweight contrast prediction coding for feature extraction of radar emitter signals. Neural Processing Letters, 2024, 56 (6): 1- 13.
18	YANG Y, GAO F, LI G Y, et al Deep learning-based downlink channel prediction for FDD massive MIMO system. IEEE Communications Letters, 2019, 23 (11): 1994- 1998. doi: 10.1109/LCOMM.2019.2934851
19	TU Y, LIN Y, HOU C B, et al Complex-valued networks for automatic modulation classification. IEEE Trans. on Vehicular Technology, 2020, 69 (9): 10085- 10089. doi: 10.1109/TVT.2020.3005707
20	GOPALAKRISHNAN S, CEKIC M, MADHOW U. Robust wireless fingerprinting via complex-valued neural networks. Proc. of the IEEE Global Communications Conference, 2019. DOI:10.1109/GLOBECOM38437.2019.9013154.
21	XIAO F, HUANG Y, ZUO Y Y, et al Over-the-air adversarial attacks on deep learning Wi-Fi fingerprinting. IEEE Internet of Things Journal, 2023, 10 (11): 9823- 9835. doi: 10.1109/JIOT.2023.3236314
22	DUARTE G J, PEREIRA T A, NASCIMENTO E J, et al. How do loss functions impact the performance of graph neural networks? Proc. of the Brazilian Congress of Computational Intelligence, 2021. DOI: 10.21528/cbic2021-161.
23	CHEN J F, ZHAO J C, HAN J, et al. Softmax clustering algorithm based on deep features representation. Journal of Nanjing University, 2020, 56(4): 533−540. (in Chinese)
24	WEN Y D, ZHANG K P, LI Z F, et al. A discriminative feature learning approach for deep face identification. Proc. of the European Conference on Computer Vision, 2016. DOI: 10.1007/978-3-319-46478-7_31.
25	WANG F, XIANG X, CHENG J, et al. Normface: L2 hypersphere embedding for face verification. Proc. of the 25th ACM International Conference on Multimedia, 2017. DOI: 10.48550/ARXIV.1704.06369.
26	WANG F, LIU W Y, LIU H J, et al. Additive margin softmax for face verification. IEEE Signal Processing Letters, 2018, 25(7): 926−930.
27	LIANG X Z, WANG X B, LEI Z, et al. Soft-margin softmax for deep classification. Proc. of the 24th International Conference on Neural Information Processing, 2017: 413–421.
28	HAN G J, CHEN C K, XU Z X, et al. Sphere margins softmax for face identification. Proc. of the 39th Chinese Control Conference, 2020. DOI: 10.23919/ccc50068.2020.9188526.
29	DENG J K, GUO J, YANG J, et al ArcFace: additive angular margin loss for deep face recognition. IEEE Trans. on Pattern Analysis and Machine Intelligence, 2022, 44 (10): 5962- 5979. doi: 10.1109/TPAMI.2021.3087709
30	CHENG X, XIE T S, WANG X M, et al Selective output smoothing regularization: regularize neural networks by softening output distributions. Applied Intelligence, 2025, 55, 721. doi: 10.1007/s10489-025-06539-6
31	LIU W Y, WEN Y D, YU Z D, et al. Sphereface: deep hypersphere embedding for face identification. Proc. of the IEEE Conference on Computer Vision and Pattern Identification, 2017. DOI: 10.1109/CVPR.2019.00482.
32	TU Y, LIN Y, ZHA H R, et al Large scale real-world radio signal recognition with deep learning. Chinese Journal of Aeronautics, 2022, 35 (9): 35- 48. doi: 10.1016/j.cja.2021.08.016
33	HE K W, ZHANG X, REN S, et al. Deep residual learning for image identification. Proc. of the IEEE Conference on Computer Vision and Pattern Identification, 2016. DOI: 10.48550/arXiv.1512.03385.
34	YU W, GUI G, HARIS G, et al An efficient specific emitter identification method based on complex-valued neural networks and network compression. IEEE Journal on Selected Areas in Communications, 2021, 39, 2305- 2317.

Algorithm	Dataset 1		Dataset 2		Number of parameters	Single-epoch time/s	Number of training epoch	Total time/s
Algorithm	Training accuracy/%	Validation accuracy/%	Training accuracy/%	Validation accuracy/%	Number of parameters	Single-epoch time/s	Number of training epoch	Total time/s
ResNet [33]	96.02	82.67	75.48	52.50	485 406	15.11	33	498.33
CVCNN- ResNet [34]	95.32	83.08	86.56	84.28	6 852 817	57.365	16	917.84
CPC-ResNet	97.35	82.77	91.32	86.67	3 951 283	52.285	15	784.275
MCPC-ResNet [17]	97.64	83.41	93.21	91.43	9 087 283	70.588	6	453.528
The proposed method	99.91	98.68	99.62	98.21	14 093 495	114.706	5	573.53

[1]	Yurui ZHAO, Xiang WANG, Liting SUN, Zhitao HUANG. Specific emitter identification based on frequency and amplitude of the signal kurtosis [J]. Journal of Systems Engineering and Electronics, 2025, 36(2): 333-343.
[2]	Rong FAN, Chengke SI, Yi HAN, Qun WAN. RFFsNet-SEI: a multidimensional balanced-RFFs deep neural network framework for specific emitter identification [J]. Journal of Systems Engineering and Electronics, 2024, 35(3): 558-574.

Embedded RF fingerprint interpretation: multi-channel complex residual networks with adaptive sphere space decision boundaries

RichHTML

PDF (PC)

Knowledge

Abstract

Cite this article

Share this article

Figures/Tables 8

References 34

Related Articles 2

Recommended Articles

Metrics

Comments