Designing cascade-resilient interdependent networks by optimum allocation of interdependencies

Many critical infrastructures are interdependent networks due to the services they receive from one another. Smart grids are examples of such systems consisting of interdependent power and cyber networks. In spite the fact that such interdependencies are essential for the operation of the whole system, they can also fuel cascade of failures within and across the networks as the failure of a service provider in one network can pose failure risk to the dependent components in the other network. Identifying optimal interdependencies to support the required services while reducing the risk of cascading failures is important in designing cascade-resilient interdependent networks. The goal of this paper is to answer: how much and from which nodes should a node request inter-network services to enhance the cascade-resiliency of the whole system? To answer this question, we adopt the influence model, a networked Markov chain framework, and present an interdependent network model to investigate the effects of interdependencies on cascading failures and characterize the optimum allocation of interdependencies. The presented interdependent network model enables modeling general interdependency scenarios by allowing multiple, directed and weighted interdependency links and capturing the inter-dependencies in a probabilistic fashion. Using this model, we show that the topology of interdependent networks and the allocation of the interdependencies largely affect their reliability and cascading failures.