PERFORMANCE ANALYSIS OF CRYPTOGRAPHIC PSEUDORANDOM NUMBER GENERATORS

Abstract

Pseudorandom Number Generators (PRNGs) are very important in cryptography and have been used in several security applications such as authentication protocols, session keys, key distribution, and keystream for a one-time pad. Producing PTRNGs is a critical issue because it should be uniform distribution, unpredictable and independent. Several PRNGs are available but the problem is do these algorithms give better performance than Riginael Algorithms? The performance of PRNGs is determined by passing a set of statistical tests which is put by National Institute of Standards and Technology (NIST). Further, the NIST has also a recommendation for random number generation using advanced encryption standard (AES) and triple data encryption standard (TDES) algorithms with counter mode (CTR) of operation. Further, several block cipher algorithms (BCAs) and alternates are present those may produce strong PRNG than NIST recommendation. Therefore, in this research work, different BCAs such as Rigindael, MARS, Rivest Cipher (RC6), Serpent, and Twofish are explored, implemented, and analyzed. After analyzing and comparing the test results, the algorithms are arranged in the following order: RC6, AES and Twofish, MARS and Serpent on 128-bit key size, Serpent and Twofish, MARS and RC6, and AES on 192-bit key size, and Twofish, AES and Serpent, MARS and RC6 on 256-bit key size. Thus, other algorithms can be used as alternative of NIST standard which indicate better or equivalent performance. This research work may help and guide the security implementers, developers, and the researchers in producing a strong key generation that is deemed necessary for cryptographic, authentication and other security applications.