A Secure and Efficient Software Random Number Generator Applicable to Internet of Things

The application of random numbers is essential in the Internet of Things (IoT), ranging from traditional data encryption functions to secure and trustworthy technologies for intelligent IoT devices. Due to their unique advantages of flexibility and convenience, software random number generators (SRNGs) are widely used in various computational applications within IoT. The research focus is on the quality, efficiency, and structural security of the output random sequences. However, there is no completely unified structural standard for SRNGs. For instance, even widely used generators, such as the Linux generator have certain deficiencies in the quality or security of their random sequence outputs. This article proposes a more secure and efficient software random number generator, namely SESRNG. First, a dual entropy pool system is constructed using a circular shift register connected to a ring aggregation pool. This system collects multiple system entropy sources in two rounds, with Shannon entropy estimation applied for online entropy estimation of the source data. Next, we utilize dual chaotic systems as extension functions to iteratively compute the entropy source data. In the designed deterministic random number generator structure, the SHA256 algorithm is used as a post-processing function to hash the key parameters of the internal state, resulting in the final random sequence. We used three well-known test suites - ENT, NIST, and the "Information Security Technology Randomness Test Methods for Binary Sequences"- to evaluate the performance of the SRNG. The results show that SESRNG can provide high-quality random numbers that meet the needs of various IoT applications.