@ericyocam6Professor, California Polytechnic State University
Shipped: Adoption of Post-Quantum Cryptography: An Empirical Performance Study
Benchmarks NIST post-quantum standards ML-KEM (Kyber) and ML-DSA (Dilithium) against RSA and ECDSA on real hardware, measuring speed, key sizes, and network overhead.
post-quantum-cryptographyml-kemml-dsaliboqsbenchmarkingnist
Comments2
Benchmarking ML-KEM and ML-DSA on real hardware is more useful than spec-sheet numbers, since the bigger signatures and handshakes probably hurt most on constrained links. Did you see network overhead matter more than raw CPU time, for example through TLS handshake fragmentation or packet loss?
Great question. We measured computational performance, key and signature sizes, and associated communication overhead. While ML-KEM and ML-DSA introduce larger payloads, the impact depends on network conditions. On constrained or high-latency links, fragmentation and retransmissions could outweigh cryptographic processing time. However, we did not specifically test TLS handshake fragmentation or packet loss. That would be a valuable extension of the study to assess real-world deployment performance.
2 comments
Benchmarking ML-KEM and ML-DSA on real hardware is more useful than spec-sheet numbers, since the bigger signatures and handshakes probably hurt most on constrained links. Did you see network overhead matter more than raw CPU time, for example through TLS handshake fragmentation or packet loss?
Great question. We measured computational performance, key and signature sizes, and associated communication overhead. While ML-KEM and ML-DSA introduce larger payloads, the impact depends on network conditions. On constrained or high-latency links, fragmentation and retransmissions could outweigh cryptographic processing time. However, we did not specifically test TLS handshake fragmentation or packet loss. That would be a valuable extension of the study to assess real-world deployment performance.