The Cryptography and Security Engineering Research Lab (CSE, Advisor: Seo SeogChung) at our university, along with the Institute for Defense Technology Protection, part of the Glocal Lab supported by the Ministry of Education (Director: Professor Han Dong-guk), titled “Optimizing HAETAE and SMAUG-T on Cortex-M4 for Resource-Constrained IoT Devices,” has been published in the IEEE Internet of Things Journal (IF: 8.7, approximately top 6% in JCR, Q1), a world-renowned academic journal in the field of IoT.
 
This study focused on optimizing HAETAE, a digital signature algorithm selected in the Korea Post-Quantum Cryptography Contest (KpqC), and SMAUG-T, a key encapsulation algorithm, for the ARM Cortex-M4 environment, a representative 32-bit embedded processor. While these two algorithms offer smaller public data sizes—such as public keys, ciphertexts, and signatures—compared to their corresponding NIST-standard quantum-resistant cryptographic algorithms (ML-DSA and ML-KEM), giving them an advantage in IoT environments with limited communication resources, there has been a relative lack of research on high-performance implementations tailored to specific platforms, which has limited the full utilization of these advantages.

The research team corrected two implementation errors in the secret key conformity verification process of the existing HAETAE implementation and proposed a new ASNE (Autocorrelation-based Spectral-Norm Evaluation) technique that reconstructs the Fast Fourier Transform (FFT)—which was previously performed iteratively—based on autocorrelation. Through this and other Cortex-M4-specific optimizations, the team improved key generation performance by up to approximately 4.09 times compared to the existing implementation. Furthermore, in SMAUG-T, the team applied number-theoretic transformation (NTT)-based optimization to polynomial multiplication—a major bottleneck—achieving performance levels approximately 3.67 times faster for key generation, 4.49 times faster for key encapsulation, and 4.41 times faster for decapsulation compared to the existing C implementation. This has significantly improved the implementation performance of the KpqC algorithm and substantially narrowed the performance gap with NIST-standard PQC.
 
Along with these performance improvements, the research team also analyzed the impact of the small public key, ciphertext, and signature sizes of HAETAE and SMAUG-T on the communication overhead of IoT protocols. Since quantum-resistant cryptography affects overall system performance not only through computational performance but also through the transmission volume of public keys, ciphertexts, and signatures, these small public data sizes can serve as a significant advantage in environments with limited communication resources. By considering both the improved computational performance achieved through optimization and existing communication efficiency, the research team demonstrated that HAETAE and SMAUG-T can serve as competitive post-quantum cryptography (PQC) alternatives in resource-constrained IoT environments.