On the security of random sources

Abstract

Many applications rely on the security of their random number generator. It is therefore essential that such devices be extensively tested for malfunction. The purpose of a statistical test is to detect specific weaknesses in random sources. Maurer’s universal test is a very common randomness test, capable of detecting a wide range of statistical defects. The test is based on the computation of a function which is asymptotically related to the source’s entropy, which measures the effective key-size of block ciphers keyed by the source’s output. In this work we develop a variant of Maurer’s test where the test function is in theory exactly equal to the source’s entropy, thereby enabling a better detection of defects in the tested source.