Optimizing Scale-Free Network Robustness with the Great Deluge Algorithm

Abstract

This paper examines the robustness of scale-free networks against degree-based attacks on their nodes. Having a robust network means the network is less likely to suffer catastrophic failure if some parts of its structure are destroyed. Many critical real world systems such as the Internet and power grids can be modeled as networks, so finding a way to reduce the chance of failure in these systems is very important.

The robustness of these networks is increased using an optimization procedure based on edge swaps. In previous work, the optimization has been done with a simple hill climbing algorithm. The hill climber is prone to getting stuck in local optima, so Buesser et al. proposed using simulated annealing as a metaheuristic to get better solutions, getting good results.

This paper introduces a great deluge metaheuristic approach for this problem. To the author’s knowledge, this algorithm has never been used for network robustness optimization before, and shows promising results. Testing indicates that the great deluge-based optimization results in larger improvements to network robustness than the simulated annealing optimization, and is considerably faster for small networks.