Abstract
Side-channel analysis has seen rapid adoption of deep learning techniques over the past years. While many paper focus on designing efficient architectures, some works have proposed techniques to boost the efficiency of existing architectures. These include methods like data augmentation, oversampling, regularization etc. In this paper, we compare data augmentation and oversampling (particularly SMOTE and its variants) on public traces of two side-channel protected AES. The techniques are compared in both balanced and imbalanced classes setting, and we show that adopting SMOTE variants can boost the attack efficiency in general. Further, we report a successful key recovery on ASCAD(desync=100) with 180 traces, a 50% improvement over current state of the art.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Similar content being viewed by others
Notes
- 1.
Refer to the Keras API description in https://www.tensorflow.org/api_docs/python/tf/keras/preprocessing/image/ImageDataGenerator.
- 2.
- 3.
- 4.
DA(x) indicates that \(x\times 100\)% of the whole points is randomly shifted while training phase, which was suggested in [4].
References
Brier, Eric., Clavier, Christophe, Olivier, Francis: Correlation power analysis with a leakage model. In: Joye, Marc, Quisquater, Jean-Jacques (eds.) CHES 2004. LNCS, vol. 3156, pp. 16–29. Springer, Heidelberg (2004). https://doi.org/10.1007/978-3-540-28632-5_2
Benadjila, Ryad., Prouff, Emmanuel., Strullu, Rémi., Cagli, Eleonora, Dumas, Cécile: Deep learning for side-channel analysis and introduction to ASCAD database. J. Cryptogr. Eng. 10(2), 163–188 (2019). https://doi.org/10.1007/s13389-019-00220-8
Chawla, N.V., Bowyer, K.W., Hall, L.O., Kegelmeyer, W.P.: Smote: synthetic minority over-sampling technique. J. Artif. Int. Res. 16(1), 321–357 (2002)
Cagli, Eleonora., Dumas, Cécile, Prouff, Emmanuel: Convolutional neural networks with data augmentation against jitter-based countermeasures. In: Fischer, Wieland, Homma, Naofumi (eds.) CHES 2017. LNCS, vol. 10529, pp. 45–68. Springer, Cham (2017). https://doi.org/10.1007/978-3-319-66787-4_3
Chari, Suresh., Rao, Josyula R., Rohatgi, Pankaj: Template attacks. In: Kaliski, Burton S., Koç, çetin K., Paar, Christof (eds.) CHES 2002. LNCS, vol. 2523, pp. 13–28. Springer, Heidelberg (2003). https://doi.org/10.1007/3-540-36400-5_3
Coron, Jean-Sébastien, Kizhvatov, Ilya: An efficient method for random delay generation in embedded software. In: Clavier, Christophe, Gaj, Kris (eds.) CHES 2009. LNCS, vol. 5747, pp. 156–170. Springer, Heidelberg (2009). https://doi.org/10.1007/978-3-642-04138-9_12
Zaid, G., Bossuet, L., Habrard, A., Venelli, A.: Methodology for efficient CNN architectures in profiling attacks. IACR Trans. Cryptographic Hardware Embed. Syst. 1, 1–36 (2020)
Kovács, G.: smote_variants Documentation Release 0.1.0. https://readthedocs.org/projects/smote-variants/downloads/pdf/latest/, 03 February 2020
Kocher, Paul., Jaffe, Joshua, Jun, Benjamin: Differential power analysis. In: Wiener, Michael (ed.) CRYPTO 1999. LNCS, vol. 1666, pp. 388–397. Springer, Heidelberg (1999). https://doi.org/10.1007/3-540-48405-1_25
Kim, J., Picek, S., Heuser, A., Bhasin, S., Hanjalic, A.: Make some noise unleashing the power of convolutional neural netoworks for pofiled side-channel analysis. IACR Trans. Cryptographic Hardware Embed. Syst. 2019(3), 148–179 (2019)
Lee, H., Kim, J., Kim, S.: Gaussian-based SMOTE llgorithm for solving Skewe class distributions. Int. J. Fuzzy Log. Intell. Syst. 17(4), 229–237 (2017)
Standaert, François-Xavier., Malkin, Tal G., Yung, Moti: A unified framework for the analysis of side-channel key recovery attacks. In: Joux, Antoine (ed.) EUROCRYPT 2009. LNCS, vol. 5479, pp. 443–461. Springer, Heidelberg (2009). https://doi.org/10.1007/978-3-642-01001-9_26
Simard, P.Y., Steinkraus, D., Platt, J.C.: Best practices for convolutional neural networks applied to visual document analysis. In: 7th International Conference on Document Analysis and Recognition - Volume 2 (2003)
Ma, L., Fan, S.: CURE-SMOTE algorithm and hybrid algorithm for feature selection and parameter optimization based on random forests. BMC Bioinformatics 169, 2017 (2017)
Maghrebi, Houssem., Portigliatti, Thibault, Prouff, Emmanuel: Breaking cryptographic implementations using deep learning techniques. In: Carlet, Claude, Hasan, M.Anwar, Saraswat, Vishal (eds.) SPACE 2016. LNCS, vol. 10076, pp. 3–26. Springer, Cham (2016). https://doi.org/10.1007/978-3-319-49445-6_1
Picek, S., Heuser, A., Jovic, A., Bhasin, S., Regazzoni, F.: The curse of class imbalance and conflicting metrics with machine learning for side-channel evaluations. IACR Trans. Cryptographic Hardware and Embed. Syst. 2019(1), 209–237 (2019)
Wang, S., Li, Z., Chao, W., Cao, Q.: Applying adaptive over-sampling technique based on data density and cost-sensitive SVM to imbalanced learning. In: The 2012 International Joint Conference on Neural Networks (IJCNN), pp. 1–8 (2012)
Wolpert, D.H.: The lack of a prior distinctions between learning algorithms. Neural Comput. 8, 1341–1390 (1996)
Acknowledgements
We gratefully acknowledge the support of NVIDIA Corporation with the donation of the Titan Xp GPU used for this research. The authors acknowledge the support from the ‘National Integrated Centre of Evaluation’ (NICE); a facility of Cyber Security Agency, Singapore (CSA).
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Appendices
A Variants of Oversampling Techniques (SMOTE Variants)
Two approaches were already introduced in [16] (SMOTE and SMOTE-ENN). However, there are currently 85 variant of SMOTEs referring to [8]. To the best of our knowledge, the investigation for effectiveness of these schemes has not been properly conducted in terms of SCA.
The variant SMOTEs in Table 2 have developed to overcome the bias for imbalanced data for DL context. As mentioned previously, only SMOTE and SMOTE-ENN are utilized in [16]. Although the performance of SMOTE in [16] is better, many variant SMOTEs have not been utilized. Moreover, they mentioned that the role of SMOTE and SMOTE-ENN is to only increase the number of minority instance. However, in general, the oversampling techniques can be further used as compared to previous suggestion. Naturally, these techniques can be used beyond HW/HD model, because the data might be biased in practice. As such, variant SMOTEs provide benefit as preprocessing tool, which help smoothing the distribution of the data.
Moreover, as mentioned earlier, these techniques are worth investigated in the context of SCA, because there are several advantages offered by SMOTE variants, such as the change of majority and noise removal. Among 85 variant SMOTEs, we have conducted preliminary investigation on their effectiveness and only reported those who are quite successful for SCA.
B Results for All Oversampling Techniques Against AES_RD and ASCAD(desync=100)
Rights and permissions
Copyright information
© 2020 Springer Nature Switzerland AG
About this paper
Cite this paper
Won, YS., Jap, D., Bhasin, S. (2020). Push for More: On Comparison of Data Augmentation and SMOTE with Optimised Deep Learning Architecture for Side-Channel. In: You, I. (eds) Information Security Applications. WISA 2020. Lecture Notes in Computer Science(), vol 12583. Springer, Cham. https://doi.org/10.1007/978-3-030-65299-9_18
Download citation
DOI: https://doi.org/10.1007/978-3-030-65299-9_18
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-65298-2
Online ISBN: 978-3-030-65299-9
eBook Packages: Computer ScienceComputer Science (R0)