Robustness in Multilayer Networks Under Strategical and Random Attacks
Rajesh Kumar, Anurag Singh
Abstract
Many real-world complex systems are modelled by multilayer networks (MLNs) which are inter-connected and inter-dependent in nature. For example, Transport networks, Power grid networks, Biological networks etc. In these networks, damage to any network layer may affect the functionality of other network layers or the entire system of multilayer network. For example, in the Transport networks, damage to Road networks may affect the Railway network as well as Air-Transport networks. Therefore, the robustness of these networks is not only an important characteristics, but also it is a challenging area of research. In the present research work, we study the robustness behavior of synthetic MLNs under random attacks as well as targeted attack by removing the specified fraction of nodes from each network layer and compare the results with single layer networks. For both kinds of attacks, it is revealed that the single layer networks are more vulnerable towards failure as compared to the MLNs. We also study the robustness in the case of MLN where network layers contain more than one connected components. By keeping the network layers intact (not removing the nodes), we remove the inter-layer edges sequentially. It is found that before the removal of last inter-layer edge, MLN contains approximately 95% of nodes in largest connected component (LCC)in the best case and approximately 73% of nodes in (LCC) in the worst case under strategical attacks.