Evaluation of Randomness Test Results for Short Sequences

Randomness testing of cryptographic algorithms are of crucial importance to both designer and the attacker. When block ciphers and hash functions are considered, the sequences subject to randomness testing are of at most 512-bit length, "short sequences". As it is widely known, NIST has a statistical test suite to analyze the randomness properties of sequences and generators. However, some tests in this suite can not be applied to short sequences and most of the remaining ones do not produce reliable test values for the sequences in question. Consequently, the analysis method which is proposed in this suite is not suitable for evaluation of generators which produce relatively short sequences. In this work, we propose an alternative approach to analyze short sequences without tweaking the tests.