Performance Evaluation of Standardized and Candidate Post-Quantum Algorithms on Resource-Constrained Virtualized IoT
TL;DR
This study benchmarks seven Key Encapsulation Mechanisms and lattice-based schemes, particularly ML-KEM and Kyber, emerged as the most practical PQC solutions for resource-constrained IoT environments and demonstrates the usefulness of virtualized benchmarking for deployment planning.
Abstract
Quantum computing poses a significant threat to Internet of Things (IoT) communications that rely on classical cryptographic algorithms such as RSA, Diffie–Hellman, and Elliptic Curve Cryptography (ECC). These algorithms are vulnerable to Shor’s algorithm, making long-term data security uncertain, particularly for IoT devices with extended operational lifespans. Post-quantum cryptographic (PQC) algorithms offer resistance to quantum attacks; however, their deployment on resource-constrained IoT devices remains challenging due to limitations in processing power, memory, and energy consumption. This study benchmarks seven Key Encapsulation Mechanisms (KEMs) ML-KEM, Kyber, NTRU, BIKE, Classic McEliece, FrodoKEM, and sntrup761 across key generation, encapsulation, and decapsulation operations. Experiments were conducted using the liboqs library in a virtualized Linux environment running WSL2 to simulate constrained computing conditions. Results revealed substantial performance differences among the algorithms. ML-KEM-512 achieved the best overall performance, completing all operations in under 14 µs, while Kyber-512 showed similar efficiency. NTRU provided the fastest encapsulation but incurred significantly higher key-generation costs. BIKE, FrodoKEM, and Classic McEliece exhibited execution times unsuitable for real-time IoT applications. Overall, lattice-based schemes, particularly ML-KEM and Kyber, emerged as the most practical PQC solutions for resource-constrained IoT environments. The findings also demonstrate the usefulness of virtualized benchmarking for deployment planning.