On capacity planning for minimum vulnerability

This paper reports on a new approach to analyze network vulnerability, which is motivated by the robustness properties of the network. We introduce new vulnerability metrics, which are inherently robust to the topological changes and targeted attacks. The proposed vulnerability metrics are based on the evaluation of the variations in the value of network criticality, a global graph metric to quantify the robustness of a network, in the presence of topological failures. The proposed metrics are convex functions of link weights (capacities); therefore, we can construct convex optimization problems to design optimal weight sets in order to minimize the effect of failures. We apply these optimization problems on some well-known networks, in particular Abilene and Rocketfuel topologies, and through extensive simulations, we will show that by re-planning the network capacity using the provided optimization schemes, we can significantly decrease the vulnerability of a network to the failures and targeted attacks.