A robustness optimization method of network based on load entropy

Cascading failures can be a serious threat to network security because of the fact that the failure of a small number of nodes may trigger the collapse of the entire system. In order to avoid cascading failures, effective approaches are proposed to improve the heterogeneity of network. But there is no universal method to evaluate the heterogeneity of network. Additionally, it is still a challenge to optimize network robustness with quantitative parameters. This paper presents an evaluation and optimization design method based on information entropy. Load entropy is defined and taken as a parameter to measure network heterogeneity. Then a method of load entropy modeling and analysis is established and the positive correlation between network entropy and network robustness is verified by Monte Carlo simulation. Based on the previous research, we present a method of network robustness optimization design based on load entropy and use genetic algorithm to quickly find the network topology with larger entropy.