A task-based dependability model for k-ary n-cubes

Dependability (reliability and availability) modeling of k-ary n-cube architectures is addressed in this paper. The dependability model considered here is known as task-based dependability because the system working condition is specified by the task requirement. For the k-ary n-cube, we therefore compute the probability of finding a working k-ary m-cube. Due to the complexity of the problem, a structural decomposition technique is used to develop the analytical model. Two probability terms care required for computing either reliability or availability. The first term finds the probability that there are x working nodes in the system. Computation of this term for the availability analysis needs the solution of a simple Markov chain. The second term finds the probability that the x working nodes form the required subcube, called the task connection probability. A recursive expression, is developed for this. Analytical results are provided for various system configurations and task requirements. It is shown through simulation that the analytical model is quite accurate.