A Unified Methodology for Designing Hardware Random Number Generators Based on Any Probability Distribution

We propose a unified methodology for converting any probability distribution which satisfies a few simple conditions to a uniformly distributed random bit stream. The proposed methodology is based on a bit truncation scheme and can be trivially implemented by basic circuit modules. A sufficient condition is derived to determine the optimal truncation. The proposed methodology is verified by three different true random number generators. Experimental results reveal that the proposed methodology can cope with different types of probability distributions, and the generated random numbers are of high quality and good randomness.