Performance Evaluation of ML-KEM and ML-DSA Post-Quantum Algorithms in Real-World IoT Networks


Creative Commons License

İNCE C.

Sakarya University Journal of Computer and Information Sciences (Online), cilt.9, sa.4, ss.1004-1011, 2026 (Scopus, TRDizin)

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 9 Sayı: 4
  • Basım Tarihi: 2026
  • Doi Numarası: 10.35377/saucis...1874093
  • Dergi Adı: Sakarya University Journal of Computer and Information Sciences (Online)
  • Derginin Tarandığı İndeksler: Scopus, Applied Science & Technology Source, Central & Eastern European Academic Source (CEEAS), Directory of Open Access Journals, TR DİZİN (ULAKBİM)
  • Sayfa Sayıları: ss.1004-1011
  • İnönü Üniversitesi Adresli: Evet

Özet

The recent introduction of Google’s Willow quantum chip has demonstrated remarkable computational capabilities, solving problems in microseconds that would take classical computers millions of years. As quantum computers advance toward cryptographic relevance, existing encryption algorithms become vulnerable to attacks that could break them within seconds or minutes, rendering current cryptographic protections ineffective. This necessitates the establishment of post-quantum cryptography (PQC) standards to secure systems in the quantum era. This study evaluates the performance of NIST-standardized PQC algorithms ML-KEM (512, 768, 1024) and ML-DSA (2, 3, 5) on a Raspberry Pi 4 Model B, representing baseline IoT device capabilities. Using the Liboqs library, each algorithm was executed across 1000 iterations in a real-world wireless network environment connected to a router, accounting for practical network conditions such as packet loss, latency, and jitter to assess computational feasibility under resource-constrained conditions. Experimental results demonstrate the practical viability of PQC algorithms in IoT environments. ML-KEM-512 achieves encapsulation in 55.76 μs and decapsulation in 56.15 μs, with key generation in 123.78 μs (median), while ML-DSA-44 completes signature generation in 379.83 μs and verification in 159.38 μs. These findings confirm that both ML-KEM and ML-DSA algorithms offer efficient and rapid performance, making them suitable candidates for securing IoT devices against quantum threats.