Cryptographic Analysis of Random Sequences

Abstract

Cryptographically strong random sequences are essential in cryptography, digital signatures, challenge-response systems, and in Monte Carlo simulation. This thesis examines techniques for cryptographic hardening of random sequences that are not cryptographically strong. Specific random sequences that are considered include d-sequences, that is sequences that are reciprocals of primes, and a new sequence obtained by the use of a specific two-dimensional mesh array. It is shown that the use of many-to-one mapping on blocks of the raw sequence improves the quality of autocorrelation function. Various types of many-to-one mappings are used and their effect on the autocorrelation function is compared. Sequences are also compared using another measure of randomness.