Performability of a Secure Electronic Voting Algorithm

This paper considers modelling the performance and reliability of a secure electronic voting scheme. The scheme provides secure verifiable blind voting, however there is a considerable administration overhead to this level of security. A Markovian process algebra is used to build a performance model of a basic system of n distributed voters. This model is shown to suffer from the familiar state space explosion problem. A simpler model is therefore developed to allow larger and more practically relevant systems to be studied. The original model is then extended to include the possibility that voters may fail and two modes of recovery are considered. The models are evaluated numerically using data obtained from measuring an implementation of this scheme in order to determine the accuracy of the approximations.