Module‑Lattice‑Based Key‑Encapsulation Mechanism (ML‑KEM) — Performance Measurements

Abstract

Key exchange mechanisms are foundational to secure communication, yet traditional methods face challenges from quantum computing. The Module‑Lattice‑Based Key‑Encapsulation Mechanism (ML‑KEM) is a post‑quantum cryptographic key exchange protocol with no known successful quantum vulnerabilities. This study evaluates ML‑KEM using experimental benchmarks.

We implement ML‑KEM in Python for clarity and in C++ for performance, with the latter showing substantial improvements. The C++ version achieves microsecond‑level execution times for key generation, encapsulation, and decapsulation. Python, though slower, offers a user‑friendly introduction to ML‑KEM's operation.

Our Python benchmark confirmed that ML‑KEM consistently outperformed RSA in execution speed across all tested parameters. ML‑KEM leverages the hardness of the Module Learning With Errors (MLWE) problem, ensuring resilience against quantum, classical, and AI‑based attacks. For practical feasibility, we tested the C++ implementation on a Raspberry Pi 4B (IoT use case). Attempts to benchmark on ESP32/ESP8266/Pico failed due to memory constraints. Results show ML‑KEM can operate in constrained environments that have sufficient resources.

Methodology & Implementation

We implemented ML‑KEM in both Python and C++. The Python implementation focuses on algorithmic clarity and educational value, while the C++ implementation prioritizes performance and resource efficiency. Benchmarks cover key generation, encapsulation, and decapsulation across multiple parameter sets.

Results & Benchmarks

C++ achieves microsecond‑level execution across KEM operations, substantially outperforming Python as expected. In comparative tests, ML‑KEM outperformed RSA across all parameters on the Python baseline. IoT feasibility testing shows Raspberry Pi 4B can execute KEM operations reliably, while ESP32/ESP8266/Pico were constrained by memory.

Conclusion

ML‑KEM demonstrates strong performance and practicality for post‑quantum key exchange, with implementability on moderately constrained devices. Its security relies on MLWE hardness, presenting resilience to classical, quantum, and AI‑driven attacks.

References

• National Institute of Standards and Technology (NIST) Post‑Quantum Cryptography Standardization.
• Module‑LWE foundational works and Kyber/ML‑KEM specifications.

Publication