Towards true random number generation in mobile environments

Abstract

In our paper, we analyze possibilities to generate true random data in mobile devices such as mobile phones or pocket computers. We show how to extract arguably true random data with a probability distribution ∈ = 2 -64 close to the uniform distribution in the trace distance. To postprocess the random data acquired from the camera we use a randomness extractor based on the Carter-Wegman universal2 families of hashing functions. We generate the data at the bit rate approximatively 36 bits per second - we used such a low bit rate only to allow statistical testing at a reasonable level of confidence. © Springer-Verlag 2009.