Cost-constrained and centrality-balanced network design improvement

Abstract

Improving resilience against failures and targeted attacks is an important aspect of network design. The resilience and cost of networks are two opposing objectives in which a designer should consider when building networks. We develop a heuristic algorithm that balances the centrality of networks by adding a set of links that minimizes the variance of graph centrality measures in a least costly fashion. Moreover, our algorithm limits the addition of links by a budget constraint. We apply our algorithm to three different realistic topologies and measure the performance of the improved graphs in terms of flow robustness when subjected to targeted attacks. Our results indicate that degree-balanced networks are more resilient than both betweenness-balanced and closeness-balanced networks.