EMMap: A Systematic Framework for Spatial EMFI Mapping and Fault Classification on Microcontrollers

2025-12-09

AI Summary

Existing research lacks a unified, reproducible methodology for modeling electromagnetic fault injection (EMFI) sensitivity and classifying fault behaviors in microcontrollers (MCUs), particularly in quantifying how probe placement affects fault outcomes. This paper introduces the first platform-agnostic framework for spatial EMFI sensitivity mapping and multi-level fault classification. It integrates precise electromagnetic field localization, systematic fault response acquisition, and unsupervised clustering analysis to characterize spatial sensitivity distributions across MCU die surfaces, while establishing a standardized pipeline for fault behavior categorization. Evaluated on three MCUs—including the Xtensa LX6 (ESP32)—using the ChipWhisperer platform, the framework successfully identifies sensitivity hotspots and distinguishes distinct fault classes (e.g., control-flow, data corruption, timing anomalies). This work bridges a critical gap in EMFI experimental practice, significantly enhancing reproducibility, cross-study comparability, and architectural portability in embedded security evaluation.

Technology Category

• Machine Learning: Hardware-aware ML
• Computer Vision: Adversarial Attacks & Robustness
• Knowledge Representation and Reasoning: Diagnosis and Abductive Reasoning

Application Category

• Security and Privacy: Large-scale security measurements
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Systems and Infrastructure for Web, Mobile and WoT: Web performance, measurement, and characterization

Abstract

Electromagnetic Fault Injection (EMFI) is a powerful technique for inducing bit flips and instruction-level perturbations on microcontrollers, yet existing literature lacks a unified methodology for systematically mapping spatial sensitivity and classifying resulting fault behaviors. Building on insights from O'Flynn and Kuhnapfel et al., we introduce a platform-agnostic framework for Spatial EMFI Mapping and Fault Classification, aimed at understanding how spatial probe position influences fault outcomes. We present pilot experiments on three representative microcontroller targets including the Xtensa LX6 (ESP32) and two ChipWhisper boards not as definitive evaluations, but as illustrative demonstrations of how the proposed methodology can be applied in practice. These preliminary observations motivate a generalized and reproducible workflow that researchers can adopt when analyzing EMFI susceptibility across diverse embedded architectures.

Problem

• Lacks unified methodology for spatial EMFI mapping and fault classification
• Understands how spatial probe position influences fault outcomes
• Generalizes workflow for analyzing EMFI susceptibility across embedded architectures

Innovation

• Systematic spatial mapping of EMFI sensitivity
• Platform-agnostic framework for fault classification
• Generalized reproducible workflow for diverse architectures