Enhancing Robustness Of Interdependent Networks Based On A Node-protection Strategy

Recently,the property,function,and their relations of complex networks have attracted much attention.In topology,complex networks are used to represent the real systems in the form of graphs which consist of nodes and interactions between nodes.Study and optimization of complex networks,can help solve difficult problems in many engineering fields.Such as,improving transportation efficiency,and robustness of power grid.Coupled network is a special form of complex network.Failures of some nodes in one network would lead to damage of nodes' functionality in the network coupled with it,which makes coupled networks more vulnerable when being attacked.Therefore,it is important to come up with a soluton to improve the robustness of coupled networks.Most existing achievements mainly focused on the integrity properties of coupled networks.However,failures also exist when networks are being reconstructed because there are interdependent links which can spread failures,and the community of networks changes in the meantime.Moreover,existing node-protection methods which aim to enhance the robustness of coupled networks only protect the influential nodes in one layer.In this thesis two methods aim to improve the robustness of coupled networks are proposed.Which are: 1)node-protection method based on community detection and 2)two-layerprotection method based on protecting influential nodes.In the community-detection based method,community detection technique is introduced into node protection.In the two-layerprotection method,the two coupled layer are regarded as a whole and the influential nodes are selected from it and are protected.Both methods performs good.The major contribution of this thesis is mainly reflected in two aspects:1)Cascading failures are considered when networks are being reconstructed.And community property are introduced into the process of improving the robustness of coupled networks.The robustness can be improved by protecting influential nodes in communities.2)The proposed method,two-layer-protection method,is tested on three different datasets.The experimental results are compared with traditional methods and the community-detectionbased method proposed in this thesis.Both the two-layer-protection method and the comunity-detection based method have a good effect on the robustness of coupled networks.