ML Security PapersApplication-Specific Power Side-Channel Attacks and Countermeasures: A Survey
Sahan Sanjaya 1, Aruna Jayasena 2, Prabhat Mishra 1
Published on arXiv
2512.23785
Model Theft
OWASP ML Top 10 — ML05
Key Finding
Power side-channel attacks are the most researched side-channel type (36.5% of published work 2015–2020) and have expanded well beyond cryptography to include ML model architecture extraction and confidential input recovery.
Side-channel attacks try to extract secret information from a system by analyzing different side-channel signatures, such as power consumption, electromagnetic emanation, thermal dissipation, acoustics, time, etc. Power-based side-channel attack is one of the most prominent side-channel attacks in cybersecurity, which rely on data-dependent power variations in a system to extract sensitive information. While there are related surveys, they primarily focus on power side-channel attacks on cryptographic implementations. In recent years, power-side channel attacks have been explored in diverse application domains, including key extraction from cryptographic implementations, reverse engineering of machine learning models, user behavior data exploitation, and instruction-level disassembly. In this paper, we provide a comprehensive survey of power side-channel attacks and their countermeasures in different application domains. Specifically, this survey aims to classify recent power side-channel attacks and provide a comprehensive comparison based on application-specific considerations.
Key Contributions
- Comprehensive taxonomy of power side-channel attacks across four application domains: cryptographic implementations, ML model reverse engineering, user behavior data exploitation, and instruction-level disassembly
- Comparative analysis of application-specific countermeasures at hardware, circuit, and algorithm levels
- Identification of gaps in existing surveys that focus narrowly on cryptographic implementations, extending coverage to ML and other domains
🛡️ Threat Analysis
The survey explicitly covers power side-channel attacks used to reverse engineer ML model architectures and extract proprietary model parameters — a physical model theft vector distinct from API-based extraction.