Machine Learning Power Side-Channel Attack on SNOW-V

TL;DR

This paper demonstrates that SNOW-V, a 5G encryption standard candidate, is vulnerable to machine learning power side-channel attacks, with neural networks significantly reducing the number of traces needed for key recovery.

Contribution

It introduces a machine learning-based power analysis attack on SNOW-V, showing its effectiveness and highlighting the need for improved countermeasures.

Findings

FCN achieved > 5X lower traces to disclosure than CPA+LDA

Power leakage confirmed by TVLA analysis

SNOW-V is vulnerable to ML-based side-channel attacks

Abstract

This paper demonstrates a power analysis-based Side-Channel Analysis (SCA) attack on the SNOW-V encryption algorithm, which is a 5G mobile communication security standard candidate. Implemented on an STM32 microcontroller, power traces captured with a ChipWhisperer board were analyzed, with Test Vector Leakage Assessment (TVLA) confirming exploitable leakage. Profiling attacks using Linear Discriminant Analysis (LDA) and Fully Connected Neural Networks (FCN) achieved efficient key recovery, with FCN achieving > 5X lower minimum traces to disclosure (MTD) compared to the state-of-the-art Correlational Power Analysis (CPA) assisted with LDA. The results highlight the vulnerability of SNOW-V to machine learning-based SCA and the need for robust countermeasures.