Reliability ranking of communication networks

In this paper a class of algorithms are proposed which allow the comparison of a finite set of communication networks in order to choose the one which is uniformly maximally reliable, i.e. has the best reliability for any value of the link failure probability. The communication networks are modelled as directed graphs whose edges fail independently with equal probability, while the nodes are considered perfectly reliable. The developed algorithms, based on simple paths or p-acyclic subgraphys, evaluate the reliability polynomial coefficients by which it is possible to verify if a sufficient condition for the uniformly maximally reliable property holds. Several examples are presented to show the practical applications of the proposed approach.