Evaluation and comparison of several graph robustness metrics to improve network resilience

Computer networks serve as critical infrastructure to services in business, health care, and education. Targeted attacks and random failures may cause link or node removals, which in turn can cause significant disruption to the availability of network services. Designing a network topology to provide acceptable levels of service in the face of these challenges can save both lives and money. A number of graph robustness metrics have been introduced to measure network resilience against such attacks. One way to improve resilience against such challenges is to add a set of links to improve these graph robustness metrics. In this paper, we add links to a given graph to improve given robustness functions. Then, we evaluate non- and improved graphs by applying centrality-based attacks to examine their resilience. Our results show that adding links to balance link-betweenness yields the best network resilience against such attacks among the studied robustness metrics.