31th Congress of the International Council of the Aeronautical Sciences

09 - Air Transport System Efficiency


B. Cui¹, L. Meilong¹, R. Curran, Delft University of Technology, Netherlands; ¹Nanjing University of Aeronautics and Astronautics, China

Cascading failure is a common phenomenon when an attack happened in a complex network. But because most of the risks are not substantial enough, we discuss the risk diffusion model and the detailed process of cascading failure in a network model.rnWe use the CASCADE model to simulate the cascading failure process of the ER network and give a definition of the Cascading Failure of Nodes and Their Links (CFNL) phenomenon. It mainly related to the limited risk conditions. Since the nodes and their links are affected by the risk propagation threshold and their load, the risk cannot simply be considered to cause the nodes and all its links to enter the first stage described in Ref 2 immediately. We describe the network A and B are independent with their original state and presume that there is a limited risk occurred in A5 like the previous situation and caused a CFNL with seven assumptions. Since the failure will suspend, we get eleven products of the network evolution. Some nodes are isolated. Some links became unidirectional. Those figures are Shrank Networks(SN), which is the name of a shrinking state of a network, compared with its origination. We also plan to analyze the changes in the degree distribution and some other topology characteristics.rnWe also define the relationship between the risk propagation threshold and the load of the network element. We suppose the suspending reasons of CFNL are mainly the limit of links and their capacities which exist in real systems. We define that the margin of each the node and link are affected by their Connectivity and Load intensity with particular mathematical expression.rnThis method relates the risk distribution limit of the network element directly to its burden. Therefore, it is possible to further analyze the connection between network unit weight and risk spread. Therefore, the significance of the analysis of CFNL is that when all failures are not simultaneously, we can analyze the current

View Paper