Kilroy Was Here: The First Step Towards Explainability of Neural Networks in Profiled Side-Channel Analysis

Valk, Daan van der; Picek, Stjepan; Bhasin, Shivam

doi:10.1007/978-3-030-68773-1_9

Cited by 11 publications

(11 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Without considering the plaintext, our approach along with [49,62,63] have been quite effective and comparable to the works using CNN. [48] Discusses a stopping criterion based on the analysis of mutual information.…”

Section: Cnnmentioning

confidence: 77%

“…Generalization is the ability of a SCA model trained on one dataset to adapt well to new, unseen data. In this regard, the authors in [62] demonstrate attacks on different SCA datasets and show how MLPs are internally the same despite changing the device or the key because MLPs are considered universal approximators. Furthermore, even small MLP networks were shown to roughly learn the same function without overfitting and generalize to different datasets [51,62].…”

Section: The Issue Of Generalizationmentioning

confidence: 99%

“…In this regard, the authors in [62] demonstrate attacks on different SCA datasets and show how MLPs are internally the same despite changing the device or the key because MLPs are considered universal approximators. Furthermore, even small MLP networks were shown to roughly learn the same function without overfitting and generalize to different datasets [51,62]. Generalization has also been addressed in [49], where the authors show how ensembles of models based on averaged class probabilities can further improve the network generalization to various datasets.…”

Section: The Issue Of Generalizationmentioning

confidence: 99%

“…Indeed, the generalizability of InfoNEAT is further boosted by employing stacking learning to improve the generalization among various classes. However, in contrast to InfoNEAT, the models in [49,51,62] are designed on a trial-and-error basis. The tuning of the network (hyperparameters) is not even considered in most of the studies, and when they are considered, the process is not automated to reach peak SCA performance like InfoNEAT.…”

Section: The Issue Of Generalizationmentioning

confidence: 99%

See 3 more Smart Citations

InfoNEAT: Information Theory-based NeuroEvolution of Augmenting Topologies for Side-channel Analysis

Acharya¹,

Ganji²,

Forte³

2021

Preprint

View full text Add to dashboard Cite

Profiled side-channel analysis (SCA) leverages leakage from cryptographic implementations to extract the secret key. When combined with advanced methods in neural networks (NNs), profiled SCA can successfully attack even those crypto-cores assumed to be protected against SCA. Despite the rise in the number of studies devoted to NN-based SCA, existing methods could not systematically address the challenges involved in the NN-based SCA. A range of questions has remained unanswered, namely: how to choose a NN with an adequate size, how to tune the NN's hyperparameters, when to stop the training, and how to explain the performance of the NN model in quantitative terms, in the context of SCA. Our proposed approach, "InfoNEAT, " tackles these issues in a natural way. InfoNEAT relies on the concept of evolution of NNs (both the network architecture and parameters, so-called neuroevolution), enhanced by information-theoretic metrics to guide the evolution, halt it with a novel stopping criteria, and improve time-complexity and memory footprint. The performance of InfoNEAT is evaluated by applying it to publicly available datasets composed of real side-channel measurements. In addition to the considerable advantages regarding the automated configuration of NNs, InfoNEAT demonstrates significant improvements over other approaches including a reduction in the number of epochs and width of the NN (i.e., the number of nodes in a layer) by factors of at least 1.25 and 6.66, respectively. According to our assessment and on the basis of our results, this is indeed achieved without any deterioration in the performance of SCA compared to the state-of-the-art NN-based methods. CCS CONCEPTS• Security and privacy → Side-channel analysis and countermeasures; • Mathematics of computing → Information theory; • Computing methodologies → Ensemble methods; Neural networks;

show abstract

Section: Cnnmentioning

confidence: 77%

Section: The Issue Of Generalizationmentioning

confidence: 99%

Section: The Issue Of Generalizationmentioning

confidence: 99%

Section: The Issue Of Generalizationmentioning

confidence: 99%

See 2 more Smart Citations

InfoNEAT: Information Theory-based NeuroEvolution of Augmenting Topologies for Side-channel Analysis

Acharya¹,

Ganji²,

Forte³

2021

Preprint

View full text Add to dashboard Cite

show abstract

“…Translation-invariance property of convolutional networks can be exploited to attack desynchronized traces [ 42 , 57 ], whereas previously described attacks would require synchronization of the traces during preprocessing, e.g., by using autocorrelation, due to their univariate nature. A distinct disadvantage of using the machine-learning-based blackbox approach is limited explainability of the results [ 58 ].…”

Section: Non-profiled Attacksmentioning

confidence: 99%

A Comprehensive Survey on the Non-Invasive Passive Side-Channel Analysis

Socha

Miškovský²,

Novotný³

2022

Sensors

View full text Add to dashboard Cite

Side-channel analysis has become a widely recognized threat to the security of cryptographic implementations. Different side-channel attacks, as well as countermeasures, have been proposed in the literature. Such attacks pose a severe threat to both hardware and software cryptographic implementations, especially in the IoT environment where the attacker may easily gain physical access to a device, leaving it vulnerable to tampering. In this paper, we provide a comprehensive survey regarding the non-invasive passive side-channel analysis. We describe both non-profiled and profiled attacks, related security metrics, countermeasures against such attacks, and leakage-assessment methodologies, as available in the literature of more than twenty years of research.

show abstract

Domain‐Adaptive Power Profiling Analysis Strategy for the Metaverse

Li,

Yang,

Liu

et al. 2024

Int J Network Mgmt

View full text Add to dashboard Cite

In the surge of the digital era, the metaverse, as a groundbreaking concept, has become a focal point in the technology sector. It is reshaping human work and life patterns, carving out a new realm of virtual and real interaction. However, the rapid development of the metaverse brings along novel challenges in security and privacy. In this multifaceted and complex technological environment, data protection is of paramount importance. The innovative capabilities of high‐end devices and functions in the metaverse, owing to advanced integrated circuit technology, face unique threats from side‐channel analysis (SCA), potentially leading to breaches in user privacy. Addressing the issue of domain differences caused by different hardware devices, which impact the generalizability of the analysis model and the accuracy of analysis, this paper proposes a strategy of portability power profiling analysis (PPPA). Combining domain adaptation and deep learning techniques, it models and calibrates the domain differences between the profiling and target devices, enhancing the model's adaptability in different device environments. Experiments show that our method can recover the correct key with as few as 389 power traces, effectively recovering keys across different devices. This paper underscores the effectiveness of cross‐device SCA, focusing on the adaptability and robustness of analysis models in different hardware environments, thereby enhancing the security of user data privacy in the metaverse environment.

show abstract

Kilroy Was Here: The First Step Towards Explainability of Neural Networks in Profiled Side-Channel Analysis

Cited by 11 publications

References 11 publications

InfoNEAT: Information Theory-based NeuroEvolution of Augmenting Topologies for Side-channel Analysis

InfoNEAT: Information Theory-based NeuroEvolution of Augmenting Topologies for Side-channel Analysis

A Comprehensive Survey on the Non-Invasive Passive Side-Channel Analysis

Domain‐Adaptive Power Profiling Analysis Strategy for the Metaverse

Contact Info

Product

Resources

About