Horizontal SCA Attacks against kP Algorithm Using K-Means and PCA

TL;DR

This paper demonstrates the effectiveness of machine learning-based side channel attacks, specifically K-means and PCA, against ECDSA implementations, achieving high accuracy even with countermeasures in place.

Contribution

It introduces novel horizontal side channel attack methods using K-means and PCA, showing their success against protected cryptographic implementations.

Findings

100% attack accuracy on unprotected implementation

88.7% accuracy with bus address sequencing countermeasure

98.3% accuracy when inputs are attacker-controlled during verification

Abstract

Side Channel Analysis attacks take advantage of the information leaked from the implementations of cryptographic algorithms. In this paper we describe two key revealing methods which are based on machine learning algorithms: K-means and PCA. We performed the attacks against ECDSA implementations without any prior knowledge about the key and achieved 100% accuracy for an implementation without any countermeasures against horizontal attacks and 88.7% accuracy for an implementation with bus address sequencing. In the scenario where the kP operation inputs are controlled by the attacker (as during signature verification), we achieved 98.3% accuracy for the implementation with countermeasures.