Summary of Power Side-channel Leakage Localization Through Adversarial Training Of Deep Neural Networks, by Jimmy Gammell et al.
Power side-channel leakage localization through adversarial training of deep neural networks
by Jimmy Gammell, Anand Raghunathan, Kaushik Roy
First submitted to arxiv on: 29 Oct 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Cryptography and Security (cs.CR)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This paper proposes a technique to identify which time steps in power traces are responsible for leaking cryptographic keys, by playing an adversarial game between a deep learning-based side-channel attacker and a trainable noise generator. The attacker aims to classify sensitive variables from the recorded power traces during encryption, while the noise generator introduces minimal noise to thwart this attack. Experimental results on synthetic datasets show that our method outperforms existing techniques in the presence of common countermeasures like Boolean masking and trace desynchronization. Although real-world results are limited due to hyperparameter sensitivity and lack of a holdout dataset, we believe this work represents an important step towards deep side-channel leakage localization without strong assumptions. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper is about using computer algorithms to find which parts of a power signal are giving away secrets, like a password. The bad guys want to steal the secret by looking at the power signal, but the good guys want to stop them by adding some noise to the signal. They play a game where they try to figure out what’s happening in the signal, and the good guys add just enough noise to confuse them. It works really well on fake data, but it doesn’t work as well on real data because there are too many things that can go wrong. |
Keywords
» Artificial intelligence » Deep learning » Hyperparameter
