Skip to content
Open access

Deep-learning side-channel analysis of embedded ML-KEM: Single-trace message recovery, cross-device generalization, and frequency-band leakage characterization

Nov 2026 · PLoS ONE · Vol 21, pp. e0358974 · 0 citations · 41 references
Medicine

TL;DR

This study evaluates ML-KEM on an ARM Cortex-M4 microcontroller under deep learning(DL)-based side-channel attacks (DL-SCA), revealing that a single power trace enables 32-byte message recovery with 99.98% accuracy, while cross-device testing across three chips achieves over 99% accuracy, highlighting critical DL-driven threats.

Abstract

Quantum computing undermines traditional public-key cryptosystems like RSA and ECC, prompting NIST to standardize Module Lattice Key Encapsulation Mechanism (ML-KEM) as a core Post-Quantum Cryptography (PQC) scheme in August 2024. Despite its theoretical security, ML-KEM’s practical vulnerability in embedded systems remains underexplored. This study evaluates ML-KEM on an ARM Cortex-M4 microcontroller under deep learning(DL)-based side-channel attacks (DL-SCA), revealing that a single power trace enables 32-byte message recovery with 99.98% accuracy, while cross-device testing across three chips achieves over 99% accuracy, highlighting critical DL-driven threats. To better understand the spectral characteristics exploited by deep-learning-based side-channel attacks, we conduct a controlled frequency-band sensitivity study using Butterworth and Chebyshev low-pass filters applied to recorded traces. Across different cutoff frequencies and filter orders, the validation accuracy decreases substantially under aggressive attenuation, indicating that a significant fraction of exploitable leakage resides in higher-frequency components. Rather than claiming that offline digital filtering constrains a realistic adversary, we interpret these results as evidence for where leakage-sensitive information is concentrated, thereby providing guidance for implementable countermeasures such as masking, shuffling, desynchronization, and acquisition-path power/EM conditioning in embedded PQC systems. This cost-efficient solution requires no algorithm/hardware modifications, emphasizing the urgency of implementation-level protections for secure PQC deployment in resource-constrained systems.

Read PDF

Similar papers

Open access Oct 2026

A Template attack for decryption key recovery on the NTT accelerator of PQC ML-KEM

Quantum computers threaten traditional cryptography, making the development of quantum-secure cryptographic schemes essential. Module-Lattice–based Key-Encapsulation Mechanism (ML-KEM), recently standardized by NIST for post-quantum public-key encryption, plays a critical role in enabling quantum-resistant key establis...

Munkhbaatar Chinbat, Li-Ji Wu, Xiang-Min Zhang et al. · 0 citations
Book Open access Nov 2026

Masked NTT-Multiplication with Fresh Randomness on ML-KEM Decapsulation to Hinder Profiling-based Attacks for Cortex-M4

NIST standardized CRYSTALS-Kyber as an ML-KEM, a post-quantum key encapsulation mechanism to resist attacks using quantum computers. However, side-channel analysis, which exploits analog leakage from electronic devices, has already broken several implementations. In particular, the Fujisaki-Okamoto transform requires d...

Haruka Hirata, Daiki Miyahara, Kazuo Sakiyama et al. · 0 citations
Review Open access 2026

Deep Learning-Based Side-Channel Analysis of Lightweight Block Ciphers: A Survey and Gap Analysis Against the AES Baseline

Deep learning has advanced profiled side-channel analysis, but for symmetric block ciphers the progress has concentrated almost entirely on the Advanced Encryption Standard (AES). Lightweight block ciphers are now widely deployed in Internet-of-Things and edge devices and have become the subject of formal standardizati...

Boohyun An, Ahmed S. Al Ali, Cédric Tavernier et al. · 0 citations
Book Open access Nov 2026

Single-Trace SAT-Assisted Side-Channel Attack on MILENAGE

An attack that can recover the complete subscriber key from a single power trace with 40% success probability on average is presented, which combines profiled deep learning-based power analysis with Boolean satisfiability (SAT) solving.

Elena Dubrova, Sönke Jendral, Linus Backlund · 0 citations
Book Open access Nov 2026

Too Small to Hide: Single-Trace Key Recovery from ML-KEM Key Generation

It is shown that one power trace of the optimized pqm4 implementation on an Arm Cortex-M4 suffices, and that the two operations leak complementary information, and that the two operations leak complementary information.

Vahid Jahandideh · 0 citations
Conference

Securing Post-Quantum Cryptography: Mitigating Side-Channel Vulnerabilities in the Quantum Era

Post-quantum cryptographic (PQC) algorithms, designed to defend against quantum threats, present new implementation challenges, particularly their susceptibility to side-channel attacks (SCAs). Unlike traditional attacks that exploit mathematical weaknesses, SCAs target physical properties—such as timing, power, and el...

Cate Richards, Chris Woods · 0 citations

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.