Swarm in EM Hay: Particle Swarm-guided Probe Placement for EM SCA
summary
Automated probe placement uses particle swarm optimization guided by matrix-based Rényi entropy and mutual information to find EM leakage on an 11×11 grid, tested on a microcontroller and two FPGAs with ChipWhisperer hardware and Husky oscilloscope. Nyström approximation with ~100 landmarks accelerates MI estimation so the swarm runs in real time. PSO with small swarms finds high-leakage points in ~30k traces versus ~242k for full grid and cuts traces needed for CPA key recovery by up to 16× fewer traces.
AI-written summary · about
chapters (11)
- 0:00EM side-channel attacks
- 1:15Traditional probe placement
- 2:42Experimental setup and TVLA
- 3:24Challenges in EM probing
- 3:45Particle Swarm Optimization
- 5:53Matrix-based Rényi entropy
- 8:45Advantages of matrix-based entropy
- 9:27Evaluating entropy with synthetic data
- 10:52Entropy and mutual information maps
- 11:37Performance vs Sniffer
- 12:19Conclusion and future work
related talks
30mAffordable EMFI Attacks Against Modern IoT Chips
29mTowards Practical Tools for Side Channel Aware Software Engineering...
35mTalk by Adria Gascon (Google)
45mIntegrated Circuit Offensive Security
22mRADOPA: Robustifying a Design Against Optical Probing Attacks
44mIC Surgery: getting to the heart of the problem with the smallest scalpel
38mNext-Gen Detection: Harnessing LLMs for Sigma Rule Automation
16mIMap: Fast and Scalable In-Network Scanning with Programmable Switches