Reality Check on Side-Channels: Lessons learnt from breaking AES on ARM Cortex-A72...
summary
Evaluates EM side‑channel attacks on an ARM Cortex-A72 (Raspberry Pi 4B) using an EM probe + LCRO oscilloscope, TVLA/SNR for leakage checks, and CNN‑based deep learning profiles. Basic and educated (CPA with elastic alignment/threshold filtering) approaches fail; a deep learning profile recovers the AES key after about 21,696 traces, and preprocessing (filtering/realignment) cuts traces needed. Public dataset and measurement grid search are released to reproduce and improve results on Raspberry Pi 4B with deep learning.
AI-written summary · about
chapters (12)
- 0:00Side-channel attacks intro
- 1:15Real-world side-channel threats
- 2:41Side-channel analysis challenges
- 3:48ARM Cortex-A72 on Pi 4B
- 5:52Experimental setup challenges
- 8:24Leakage assessment models
- 10:32Prior side-channel research
- 11:16Basic adversary results
- 12:21Educated adversary results
- 13:05Deep learning approach
- 16:35Security trade-offs
- 17:39Conclusions and dataset
related talks
59mBreaking AES with side channel analysis
01h 02mCryptographic Side-Channels on Embedded Devices
29mTowards Practical Tools for Side Channel Aware Software Engineering...
20mBreaking the Blindfold: Deep Learning-based Blind Side-channel Analysis
41mFault Attacks & Side Channels
39mHow We Taught ChatGPT-4 to Break mbedTLS AES With Side-Channel Attacks
21mEnhanced Security for LoRaWAN devices
32mBlack box side-channel leakage verification using statistical approach