Dependence in network reliability

It is fairly easy to calculate the reliability of a network with independent nodes by using a technique called pivoting. However, when there is dependence, this calculation will prove inaccurate and a model for the dependence is required. This paper considers the issues involved in developing a suitable model. Particular emphasis is placed on ensuring the calculations and distributions involved do not become intractable for large networks. Various potential distributions are discussed. Strategies are suggested for simplifying the dependence. A model for cascading failures is proposed.