Machine Learning Enhanced Quantum-Safe Encryption: A Novel Optimisation Framework.

Area of Science:

• Cryptography and Security Engineering
• Machine Learning Applications
• Quantum Computing Impact

Background:

• NIST standardization of post-quantum cryptography (PQC) necessitates quantum-resistant algorithms.
• Increasing machine learning (ML) use demands efficient, resource-aware cryptographic primitives.
• Existing research explores ML's role in PQC optimization and privacy.

Purpose of the Study:

• To introduce QSafe-ML, a novel framework for optimizing PQC implementations using ML.
• To enhance the efficiency of NIST-standardized PQC schemes on diverse hardware.
• To ensure cryptographic performance meets quantum-safe security standards.

Main Methods:

• A four-stage framework: hardware profiling, ML-based surrogate modeling, multi-objective optimization, and security validation.
• Targeting NIST PQC lattice-based schemes (CRYSTALS-Kyber, Dilithium, Falcon, NTRU) on heterogeneous hardware.
• Utilizing repeated trials for robust experimental evaluation and ablation studies.

Main Results:

• Achieved mean latency reductions of 27.5-41.9% across platforms.
• Demonstrated memory savings of 13.3-30.2% and energy savings of 22.8-38.2%.
• All optimized configurations maintained ≥128-bit post-quantum security; surrogate-guided search was key.

Conclusions:

• QSafe-ML effectively optimizes PQC implementations for resource-constrained environments.
• The framework offers significant performance gains while upholding stringent security requirements.
• Open-sourced resources promote reproducibility in ML-assisted cryptographic system evaluation.