Modeling link weights in backbone networks

Complex networks have become a large area of study due to their pervasiveness in today's society and our dependence on them. While network science enables engineers and scientists to improve resilience of networks, due to lack of realistic data, simplistic assumptions may result in incorrect conclusions. Network service providers optimize their designs and plan for future capacities based on realistic population estimates. In this paper, we propose two synthetic network models that assign weight to links based on realistic population data of node locations. First, the weighted link model averages the population of cities to assign link weights, whereas the tiered weighted link model assigns preset link weights based on the quartile population that the link weight falls into. We study the performance of model networks under targeted attack scenarios, and our results indicate performance varies based on attack scenario.