Accelerating CKKS Homomorphic Encryption Using Minimum Modulus Precomputation Method
International Journal on Advanced Science, Engineering and Information Technology · 27.06.2026 · 16 (3) · 1019-1025
DOI: 10.18517/ijaseit.16.3.13889 ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal on Advanced Science, Engineering and Information Technology, online erschienen am 27.06.2026. Verfasst von So-Yeon Kwon und 2 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Cheon–Kim–Kim–Song (CKKS) scheme enables approximate homomorphic computation over real and complex numbers and is widely used in machine learning and data analytics applications. However, conventional CKKS implementations adopt an oversized initial ciphertext modulus to accommodate iterative modulus reduction during rescaling, resulting in increased integer arithmetic complexity and memory usage. This paper proposes a minimum modulus precomputation method that determines the minimum required initial modulus bit length, denoted as , based on the maximum input magnitude and target precision. The method derives upper bounds on ciphertext coefficient growth and cumulative error under repeated multiplication and rescaling. Two stability conditions—decryption correctness and wrap-around safety—are formulated to validate the reduced modulus configuration. Unlike the conventional approach, the proposed method maintains a fixed minimum modulus and applies shift-based rescaling without progressive modulus reduction. Theoretical analysis confirms that the reduced modulus satisfies both stability conditions. Experimental evaluations on repeated multiplication, high-degree polynomial functions, and machine learning-based operations demonstrate that the proposed method reduces the modulus size by up to 30.1% and achieves execution time reductions of up to 39.1%, while maintaining relative errors below the 1% tolerance threshold. In addition, RLWE security analysis shows that all reduced-modulus configurations preserve at least 163-bit security. These results indicate that the initial modulus in CKKS can be systematically optimized without compromising computational stability, numerical precision, or security. The proposed method can be extended to incorporate bootstrapping and adaptive modulus allocation for deeper homomorphic computations.
Zitieren
APA
Kwon, S., Ro, S., & Lee, K. (2026). Accelerating CKKS Homomorphic Encryption Using Minimum Modulus Precomputation Method. International Journal on Advanced Science, Engineering and Information Technology, 16(3), 1019-1025. https://doi.org/10.18517/ijaseit.16.3.13889
Vancouver
Kwon S, Ro S, Lee K. Accelerating CKKS Homomorphic Encryption Using Minimum Modulus Precomputation Method. International Journal on Advanced Science, Engineering and Information Technology. 2026;16(3):1019-1025. doi:10.18517/ijaseit.16.3.13889
BibTeX
@article{Kwon2026,
author = {Kwon, So-Yeon and Ro, Seong-Jun and Lee, Kwangjae},
title = {Accelerating CKKS Homomorphic Encryption Using Minimum Modulus Precomputation Method},
journal = {International Journal on Advanced Science, Engineering and Information Technology},
year = {2026},
volume = {16},
number = {3},
pages = {1019--1025},
doi = {10.18517/ijaseit.16.3.13889}
}
EUVIMED