Network Robustness Optimization And Analyses Based On Evolutionary Algorithms

Networks are very important in the real world.Not only the Internet network,transportation,human relations but also the trade,biological genes,etc.,can be abstracted as networks.These networks play extremely important roles,and their stable and efficient work are of great significance to society.Therefore,these networks are often abstracted as complex network models,and are analyzed and studied by using complex network theory.However,due to the complexity of the actual environment and the fluctuation of the system,those networks are very fragile under attaks.Therefore,it is a very urgent and important task to study the robustness of these complex networks.The research of network robustness is to study the ability of the network system to continue to operate efficiently under the condition of various attacks.The research of network robustness is very important both in the theory of complex network and in the practical applications.According to the actual situation,from the view of optimizing the topology of networks,this thesis makes an in-depth study on the network's resistance to various types of attacks:?1?This thesis presents a new Memetic algorithm to improve the robustness of scale-free networks against target and random attacks.Research shows that most of the real-world networks are scale-free networks.Because the node degree distribution of scale-free network has the characteristics of"long tail",it has a strong ability to resist random attacks,but the ability to resist malicious attacks is weak.So at this stage,most of the researches on network robustness is mainly based on the robustness of scale-free networks against malicious attacks.In a recent study,by adjusting the scale-free network topology Schneider et al.found that networks with the?onion-like?topology have high robustness against malicious attacks.However,it is found in this thesis that the?onion-like?network is vulnerable under random attacks,and the network topology that is robust under both target attack and random attack is totally different from the?onion-like?topology.In order to enhance the ability of scale-free network resisting to both malicious and random attacks,this thesis presents a new robustness measure first.Then,based on the framework of Memetic algorithm,a new evolutionary algorithm,abbreviated as MA-RSF_TRA is proposed.Through experiments on different networks,the effectiveness of the MA-RSF_TRA algorithm is verified.?2?An efficient Memetic algorithm is proposed to optimize the robustness of scale-free networks against cascading failures.In recent years,the research on cascading failures in complex networks has attracted much attention.However,most of the existing researches on cascading failures construct the cascading failure model of complex networks,most of which analyze from a theoretical point of view.Cascading failures caused by one or a few nodes are studied in these studies.The study of these theories may not be practical in real networks.In order to solve the above problems,this thesis presents a more practical evaluation criteria(R_CF)to measure the robustness of complex networks in cascading failures.Then,based on the memetic algorithm framework,a new intelligent optimization algorithm,abbreviated as MA-R_CF,is proposed to optimize the ability of complex networks against cascading failures.In MA-R_CF,a new local search operator is proposed,and R_CF is used as the objective function of the optimization algorithm.Finally,the effectiveness of the MA-R_CF algorithm is tested by using both simulated and real networks.The experimental results show that the MA-R_CF algorithm is more efficient than other common algorithms in improving the ability of complex networks to resist cascading failures.?3?A Memetic algorithm is proposed to improve the communication efficiency and the robustness of scale-free networks.Malicious attacks occur frequently in a variety of infrastructure networks,and are an important threat to network security.In the past ten years,researchers have made a deep research on malicious attacks,and some strategies improving the robustness under malicious attacks have been proposed.The communication efficiency of network is the efficiency of the exchange of information in the network.Based on the pursuit of robustness and the pursuit of efficient networks,this thesis presents an efficient evolutionary algorithm to optimize the communication efficiency of network and to enhance network ability resisting malicious attacks,abbreviated as MA-RSF_CE.The effectiveness of the MA-RSF_CE algorithm is verified by experiments on both simulated and real networks.At the same time,the systematic analysis and research on the robust and efficient network topology shows that MA-RSF_CE algorithm is of great significance for the optimization of the real network.