Pseudorandom Number Generation in Smart Cards: An Implementation, Performance and Randomness Analysis

Smart cards rely on pseudorandom number generators to provide uniqueness and freshness in their cryptographic services i.e. encryption and digital signatures. Their implementations are kept proprietary by smart card manufacturers in order to remain competitive. In this paper we look at how these generators are implemented in general purpose computers. How architecture of such generators can be modified to suit the smart card environment. Six variations of this modified model were implemented in Java Card along with the analysis of their performance and randomness. To analyse the randomness of the implemented algorithms, the NIST statistical test suite is used. Finally, an overall analysis is provided, that is useful for smart card designers to make informed decisions when implementing pseudorandom number generators.