High-Efficiency Side-Channel Analysis: From Collaborative Denoising to Adaptive B-Spline Dimension Reduction

LUO Yuling; XU Haiyang; OUYANG Xue; FU Qiang; QIN Sheng; LIU Junxiu

doi:10.11999/JEIT251047

Article Contents

Article Navigation > Journal of Electronics & Information Technology > 2025 >

LUO Yuling, XU Haiyang, OUYANG Xue, FU Qiang, QIN Sheng, LIU Junxiu. High-Efficiency Side-Channel Analysis: From Collaborative Denoising to Adaptive B-Spline Dimension Reduction[J]. Journal of Electronics & Information Technology. doi: 10.11999/JEIT251047

Citation:

LUO Yuling, XU Haiyang, OUYANG Xue, FU Qiang, QIN Sheng, LIU Junxiu. High-Efficiency Side-Channel Analysis: From Collaborative Denoising to Adaptive B-Spline Dimension Reduction[J]. Journal of Electronics & Information Technology. doi: 10.11999/JEIT251047

Citation:

LUO Yuling, XU Haiyang, OUYANG Xue, FU Qiang, QIN Sheng, LIU Junxiu. High-Efficiency Side-Channel Analysis: From Collaborative Denoising to Adaptive B-Spline Dimension Reduction[J]. Journal of Electronics & Information Technology. doi: 10.11999/JEIT251047

PDF( 8080 KB)

High-Efficiency Side-Channel Analysis: From Collaborative Denoising to Adaptive B-Spline Dimension Reduction

doi: 10.11999/JEIT251047 cstr: 32379.14.JEIT251047

LUO Yuling^{1, 2},
XU Haiyang^{1, 2},
OUYANG Xue^{1, 2
,
,},
FU Qiang^{1, 2},
QIN Sheng^{1, 2},
LIU Junxiu^{1, 2}

1.
School of Electronic and Information Engineering/College of Integrated Circuits, Guangxi Normal University, Guilin 541004, China
2.
Guangxi Key Lab of Brain-inspired Computing and Intelligent Chips, Guilin 541004, China

Funds: The National Natural Science Foundation of China (62462009, 42501519),The Guangxi Science and Technology Program Project: Guangxi Science and Technology Base and Talent Special Project (Gui Ke AD24010047), The Guangxi Young Science and Technology Talent Project (GXYESS2025144)

Accepted Date: 2025-12-22
Rev Recd Date: 2025-12-22

Available Online: 2026-01-04

Abstract

Abstract

Objective The performance of side-channel attacks is often limited by the low signal-to-noise ratio of raw power traces, the masking of local leakage by redundant high-dimensional data, and the reliance on empirical parameters during preprocessing. Existing research usually optimizes individual stages such as denoising or dimensionality reduction in isolation, lacks a systematic framework, and struggles to balance signal-to-noise ratio improvement with the preservation of local leakage features. This study proposes a unified analysis framework that integrates denoising, adaptive parameter selection, and dimensionality reduction while preserving local features. By coordinating these components, the framework effectively improves both the efficiency and robustness of side-channel attacks. Methods Leveraging the similarity of power traces corresponding to identical plaintexts and the local characteristics of B-splines, this paper presents a side-channel analysis method based on adaptive B-spline dimensionality reduction and collaborative denoising. First, a Collaborative Denoising Framework (CDF) is constructed, in which high-quality traces are selected using a plaintext-mean template, and targeted denoising is performed by SVD guided by a singular-value template. Second, the Neighbourhood Asymmetry Clustering (NAC) method is employed to adaptively determine key thresholds in the CDF. Finally, an Adaptive B-spline Dimensionality Reduction (ABDR) algorithm is proposed, which allocates nodes non-uniformly according to the trace variance profile, achieving efficient data compression while preserving local leakage characteristics. Results and Discussions Experiments on two datasets based on 8-bit AVR (OSR2560) and 32-bit ARM Cortex-M4 (OSR407) architectures show that CDF can significantly improve the signal-to-noise ratio, by 60% on OSR2560 (Figure 2) and by 150% on OSR407 (Figure 4). In addition, the number of traces required for key recovery is successfully reduced from 3000/2400 to 1200/1500 (Figures 3 and 5). By adaptively selecting the key thresholds in CDF, the proposed neighborhood asymmetry clustering (NAC) achieves faster and more stable guessing-entropy convergence than fixed-threshold and K-means strategies, thereby enhancing the robustness of the framework (Figure 6). ABDR can densely place knots in leakage regions with large variance and sparsely place knots in regions with small variance. While ensuring a high attack success rate, it compresses the data dimensionality from 5000 and 5500 to 1000 and 500, respectively, achieving a compression rate of approximately 80% and striking a balance between accuracy and efficiency. Moreover, at the optimal dimensionality (Figure 7), the correlation coefficient of the correct key is significantly higher than that of other dimensionality-reduction methods, reaching 0.1860 and 0.3605 on the OSR2560 and OSR407 datasets, respectively, which demonstrates the best local information retention capability and attack efficiency (Tables 3 and 4). Conclusions This study confirms that the proposed CDF significantly improves the signal-to-noise ratio of power traces. NAC enables adaptive parameter optimization, thereby enhancing robustness. Meanwhile, through precise local modeling, ABDR successfully mitigates the inherent trade-off between reducing high-dimensional data and preserving critical local leakage. Comprehensive experimental results demonstrate that this integrated framework effectively addresses challenges such as low signal-to-noise ratio, the masking of local information by redundant high-dimensional data, and the reliance on empirical parameters. It also offers a practical and scalable solution for side-channel analysis in real-world scenarios.

FullText(HTML)

References(22)

References

[1]	ARPAIA P, CAPUTO F, CIOFFI A, et al. Uncertainty analysis in cryptographic key recovery for machine learning-based power measurements attacks[J]. IEEE Transactions on Instrumentation and Measurement, 2023, 72: 1006108. doi: 10.1109/TIM.2023.3284933.
[2]	PANTOJA J J, BUCHELI V A, and DONALDSON R. Electromagnetic side-channel attack risk assessment on a practical quantum-key-distribution receiver based on multi-class classification[J]. EPJ Quantum Technology, 2024, 11(1): 78. doi: 10.1140/epjqt/s40507-024-00290-6.
[3]	MAJI S, BANERJEE U, and CHANDRAKASAN A P. Leaky nets: Recovering embedded neural network models and inputs through simple power and timing side-channels—Attacks and defenses[J]. IEEE Internet of Things Journal, 2021, 8(15): 12079–12092. doi: 10.1109/JIOT.2021.3061314.
[4]	RAEKER-JORDAN N, CHUNG J, KONG Zhenyu, et al. Ensuring additive manufacturing quality and cyber–physical security via side-channel measurements and transmissions[J]. Journal of Manufacturing Systems, 2024, 73: 275–286. doi: 10.1016/j.jmsy.2024.02.005.
[5]	LIU A, WANG An, SUN Shaofei, et al. CL-SCA: A contrastive learning approach for profiled side-channel analysis[J]. IEEE Transactions on Information Forensics and Security, 2025, 20: 5109–5122. doi: 10.1109/TIFS.2025.3570123.
[6]	BAO Chongxi and SRIVASTAVA A. Reducing timing side-channel information leakage using 3D integration[J]. IEEE Transactions on Dependable and Secure Computing, 2019, 16(4): 665–678. doi: 10.1109/TDSC.2017.2712156.
[7]	WU Lichao, PERIN G, and PICEK S. Weakly profiling side-channel analysis[J]. IACR Transactions on Cryptographic Hardware and Embedded Systems, 2024, 2024(3): 707–730. doi: 10.46586/tches.v2024.i3.707-730.
[8]	YOU Chunheng, CHIANG C H, CHAO P C P, et al. New adaptive template attacks against Montgomery-ladder-based ECCs in IoT devices[J]. IEEE Internet of Things Journal, 2024, 11(12): 22716–22725. doi: 10.1109/JIOT.2024.3384076.
[9]	WU Lichao, WEISSBART L, KRČEK M, et al. Label correlation in deep learning-based side-channel analysis[J]. IEEE Transactions on Information Forensics and Security, 2023, 18: 3849–3861. doi: 10.1109/TIFS.2023.3287728.
[10]	GAO Pengfei, SONG Fu, and CHEN Taolue. Compositional verification of first-order masking countermeasures against power side-channel attacks[J]. ACM Transactions on Software Engineering and Methodology, 2024, 33(3): 79. doi: 10.1145/3635707.
[11]	DUCHARME G R and MAURINE P. Estimating the Signal-to-Noise ratio under repeated sampling of the same centered signal: Applications to side-channel attacks on a cryptoprocessor[J]. IEEE Transactions on Information Theory, 2018, 64(9): 6333–6339. doi: 10.1109/TIT.2018.2851217.
[12]	MOZIPO A T and ACKEN J M. Analysis of countermeasures against remote and local power side channel attacks using correlation power analysis[J]. IEEE Transactions on Dependable and Secure Computing, 2024, 21(6): 5128–5142. doi: 10.1109/TDSC.2024.3370711.
[13]	OU Changhai, LAM S K, SUN Degang, et al. SNR-centric power trace extractors for side-channel attacks[J]. IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems, 2021, 40(4): 620–632. doi: 10.1109/TCAD.2020.3003849.
[14]	DEL POZO S M and STANDAERT F X. Blind source separation from single measurements using singular spectrum analysis[C]. The 17th International Workshop on Cryptographic Hardware and Embedded Systems (CHES 2015), Saint-Malo, France, 2015: 42–59. doi: 10.1007/978-3-662-48324-4_3.
[15]	ZHANG Fan, DONG Xiaofei, YANG Bolin, et al. A systematic evaluation of wavelet-based attack framework on random delay countermeasures[J]. IEEE Transactions on Information Forensics and Security, 2020, 15: 1407–1422. doi: 10.1109/TIFS.2019.2941774.
[16]	WANG Yuanzhen, ZHANG Hongxin, FANG Xing, et al. Hybrid threshold denoising framework using singular value decomposition for side-channel analysis preprocessing[J]. Entropy, 2023, 25(8): 1133. doi: 10.3390/e25081133.
[17]	LIU Songran and YI Wang. Task parameters analysis in schedule-based timing side-channel attack[J]. IEEE Access, 2020, 8: 157103–157115. doi: 10.1109/ACCESS.2020.3019323.
[18]	PENG Dehua, GUI Zhipeng, WANG Dehe, et al. Clustering by measuring local direction centrality for data with heterogeneous density and weak connectivity[J]. Nature Communications, 2022, 13(1): 5455. doi: 10.1038/s41467-022-33136-9.
[19]	PAGUADA S, BATINA L, and ARMENDARIZ I. Toward practical autoencoder-based side-channel analysis evaluations[J]. Computer Networks, 2021, 196: 108230. doi: 10.1016/j.comnet.2021.108230.
[20]	KONG Yinan and SAEEDI E. The investigation of neural networks performance in side-channel attacks[J]. Artificial Intelligence Review, 2019, 52(1): 607–623. doi: 10.1007/s10462-018-9640-4.
[21]	MAGHREBI H and PROUFF E. On the use of independent component analysis to denoise side-channel measurements[C]. The 9th International Workshop on Constructive Side-Channel Analysis and Secure Design (COSADE 2018), Singapore, Singapore, 2018: 61–81. doi: 10.1007/978-3-319-89641-0_4.
[22]	RIOJA U, BATINA L, FLORES J L, et al. Auto-tune POIs: Estimation of distribution algorithms for efficient side-channel analysis[J]. Computer Networks, 2021, 198: 108405. doi: 10.1016/j.comnet.2021.108405.