Increase or Decrease Network Robustness with Genetic algorithms : A method for maximization or minimization of network robustness in attack or random failure scenarios

Abstract

Network robustness is a fundamental measure for finding the network tolerance against failures and attacks. There are many methods to measure network robustness for a variety of networks like a network of routers, transportation network and so on. Increasing network robustness against failures and attacks is a fundamental problem which many methods have been introduced like random preferential node attachment, random link attachment and etc. In this paper, we present a method to increase the attack impact or decrease random failure impact on the network depending on the purpose. Following method uses genetic algorithm as an optimization approach for improving network robustness measurement function. In order to do that, we start to find a sequence of node removals which have the greatest impact on the robustness measurement function. In case of increasing the network robustness, we duplicate the aforementioned nodes. This sequence can also serve us as a guidance for attacking harmful networks, like fire or disease distribution with minimal cost.