High-speed implementation of MultiMCW entropy estimation method

Abstract

The National Institute of Standards and Technology (NIST) recently published the SP 800-90B recommendation for estimating the entropy of an entropy source and provided a Python-based estimator tool. For estimating entropy, NIST SP 800-90B is first executed to verify whether the entropy source is independent and identically distributed (IID). Then, an entropy source determined to be IID is estimated using the most common value (MCV) estimator in the IID track, whereas an entropy source determined to be non-IID is estimated using 10 estimators including the MCV estimator in the non-IID track. This study proposes two high-speed implementation methods for the MultiMCW estimator that requires the most computational time in NIST's code. We implement Python and C codes for this estimator and compare them through experiments. The results show that our proposed Python and C codes are approximately five and 700 times faster than NIST's code, respectively.