POLARIS: Explainable Artificial Intelligence for Mitigating Power Side-Channel Leakage

Power side-channel attacks pose real security risks to sensitive systems: Microelectronic devices used in manufacturing, energy, and defense can leak encryption keys and other sensitive data through their power consumption patterns, representing a significant vulnerability in critical infrastructure.

AI can help protect against these attacks: POLARIS uses explainable artificial intelligence (XAI) to automatically identify and mask the parts of a system most vulnerable to power leakage, offering a more efficient approach than previous methods like VALIANT.

The solution is practical and scalable: The framework reduces information leakage while also improving execution time and reducing computational overhead, making it feasible for deployment in real-world large-scale electronic systems.

Automation reduces human burden: POLARIS uses an unsupervised process to automatically build training datasets, meaning security engineers don't need to manually identify vulnerabilities—the AI does this work systematically.

Explainability matters for security: By using XAI rather than a "black box" approach, POLARIS allows security professionals to understand why certain parts of a system are being protected, enabling better oversight and trust in the security solution.