Research On Complex Network Overload Behavior Under Two Targeted Attack Strategies

With the development of science and technology,many systems in the real world can be modeled as complex networks.In real life,overload cascading failure events caused by node load exceeding its maximum load often occur.This brings many inconveniences and even catastrophic consequences to people's daily lives.How do different attack strategies affect the network's overload failure behavior?What effect does the interdependence between networks have on the overload failure mechanism?These are the key issues in the study of overload cascading failure of complex networks.Research on these issues can help people understand the rules of overload cascading failures in actual systems,and provide guidance for building highly robust infrastructures.Based on the above questions,the main content of this dissertation is as follows:Based on the Motter-Lai model,by using betweenness to measure node load,we study the failure behaviour of a single Erd(?)s-Rényi network and Scale-Free network due to node overload under two targeted attack strategies(Maximum degree-based strategy and Failure probability function-based strategy).Comparing the average betweenness of nodes with the same degree before and after the attack,it is found that the average betweenness growth rate of nodes with a smaller degree value is higher than that of nodes with a larger degree,and they are more likely to cause cascading failures due to overload.Therefore,the initial degree of the node is an important factor in determining its betweenness and overload failure.In addition,it is found that the system occurs a first-order phase transition at the critical thresholdp_t.This shows that the removal of even one node near the critical point can cause a complete collapse in the network.Regarding the influencing factors of the critical threshold,it is found that increasing the node capacity,the average degree of the ER network or the power-law exponent of the SF network helps to improve the robustness of the system.The above conclusions are applicable to ER network and SF network,as well as the two targeted attack strategies.In particular,for the two networks under the Failure probability function-based strategy,when the probability parameter is reduced,the failure probability of the node with a higher degree is decreased,and the critical threshold is reduced,which promotes the improvement of the network robustness.Then the single network model is extended to interdependent networks.The overload cascading failure behaviors of interdependent ER networks and interdependent SF networks under two targeted attack strategies are studied separately.Some conclusions similar to those on a single network model are found:nodes with a smaller initial degree have a higher probability of overload failure;the proportion of giant component shows a first-order discontinuous phase transition as the attack strength changes;reducing the failure probability parameter,increasing node capacity,and improving the average degree of interdependent ER networks are all helpful to improve the robustness of interdependent networks.In particular,when the interdependent SF networks suffers Maximum degree-based attack,increasing the power-law exponent can improve the robustness of the system;while in the case of the Failure probability function-based attack,increasing the power-law exponent makes the system more vulnerable.In addition,we compare the impact of different attack strategies(including:Maximum degree-based attack,Failure probability function-based attack,random attack)on the overload failure behavior of a single network and interdependent networks.The results show that the single network and the interdependent networks are the most vulnerable under the Maximum degree-based attack strategy,followed by the Failure probability function-based attack strategy,and finally the random attack strategy.Furthermore,by comparing the overload failure behavior of a single ER(SF)network and interdependent ER(SF)networks under the same targeted attack strategy,it is found that due to the dependence between the two networks,the interdependent networks becomes more fragile.