A new dual entropy core true random number generator

The entropy produced by a conventional single chaotic map based true random number generator (TRNG) is usually limited due to the finite number of Lyapunov exponents. In this work, we present a new dual entropy core TRNG architecture which is capable of producing high levels of randomness using hardware redundancy. Mathematical models of conventional and proposed TRNG architectures have been developed for a comparative analysis of the statistical and randomness properties. Our theoretical studies showed that the proposed architecture which employs Bernoulli map as the entropy source, has an inherently symmetric probability density function with zero mean. Using a practical information metric, T-entropy, we demonstrated that the proposed architecture performs better in terms of randomness, for a wide range of control parameter values when compared to its single entropy core counterpart. A proof of concept prototype of the proposed architecture is designed and implemented using a field programmable analog array integrated circuit. Random numbers acquired from the prototype have successfully passed all NIST 800.22 statistical tests.