Survivability Analysis and Optimization of Dependent Networks Based on Natural Connectivity

Abstract

In order to more fully evaluate the anti-destructive performance of the dependent network under different attacks, three one-to-one, one-to-many, and many-to-many edge-connection methods, as well as three types of degree homology, degree heterogeneity and randomness Coupling way to construct different network models. Under four different attack methods, the natural connectivity is used as a measure of the indestructibility of the dependent network, and the dependent network models of different connected edges are compared to each other based on the intentional attack of the degree and the median when they face different attack strategies. The anti-destructive effect is greater, and the dependent network has the best anti-destructive performance when the number of attack nodes is small. In addition, by changing the coupling strength between the sub-networks, it is found that the anti-destructiveness of the dependent network also increases as the coupling strength increases. Based on natural connectivity, the concept of node dispersion was proposed to evaluate the comprehensive survivability of dependent networks, and it was found that the dependent network was the most robust when one-to-many homogeneous coupling was used.