A stochastic model is a mathematical description (of the relevant properties) of an entropy source using random variables. A stochastic model used for an entropy source analysis is used to support the estimation of the entropy of the digitized data and finally of the raw data. In particular, the model is intended to provide a family of distributions, which contains the true (but unknown) distribution of the noise source outputs. Moreover, the stochastic model should allow an understanding of the factors that may affect the entropy. The distribution of the entropy source needs to remain in the family of distributions, even if the quality of the digitized data goes down.
NIST SP 800-90B