The paper presents the results of a comprehensive study of the cryptographic properties and computational efficiency of deterministic random bit generators (DRBG) standardized in NIST SP 800-90A Rev. 1. A comparative analysis of the CTRDRBG, HASHDRBG, HMACDRBG, and DualECDRBG algorithms was conducted using various cryptographic primitives, including the AES-256 block cipher, SHA-256/SHA-512 hash functions, and elliptic curves P-256 and P-384. The dependence of generator performance on the type of underlying cryptographic primitive was investigated by measuring the quantitative indicator cycles per byte (cpb) on a 64-bit architecture. The statistical properties of the generated sequences were verified using the NIST SP 800-22 Rev. 1a test suite. It was established that CTRDRBG based on AES-256 and HMACDRBG using SHA-256 provide an optimal balance between cryptographic strength and computational speed, demonstrating the highest performance among the studied implementations. DualECDRBG, based on scalar point multiplication over elliptic curves, exhibits the highest computational complexity due to the arithmetic of elliptic-curve point groups but achieves the highest statistical quality of output sequences. The study substantiates the prospects of using elliptic-curve cryptographic primitives for the design of postquantum-resistant random bit generators. The results form a methodological basis for the reasoned selection of DRBG types according to specific cryptographic application requirements, considering computational constraints and security levels. The findings can be used to optimize key-generation protocols, design electronic signature and authentication systems, and develop cryptographic mechanisms compliant with international security standards.
Pryma et al. (Wed,) studied this question.