| Various natural or social complex systems can be abstracted as a complex network, people are surrounded by various complex networks, while frequent network failures, including natural disaster, military attack, malicious attack and the components aging are threating the survivability of complex networks, those failures have serious threated peoples' lives and properties, thus, in rencent years, the grand theoretical and practical values of survivability and recovery technologies for weighted complex network have become an increasingly prominent, and have become a very promisingly leading-edge research topic. Most of existing literatures on complex networks are simplified modeling of real network. While in real world networks, the carrying capacities on different links and the processing powers of different nodes are variable. Besides, the importances and roles of links in networks also are different. Based on those realities, in the research area of complex networks, there are several issues that must be resolved:first, the caracteristics of weighted network need to be investigated deeply. Because the network weights can reflect on the difference between components, in order to have an accurate understanding, researchers should have more indepth study on weighted network. Second, the robustness and survivability of weighted networks need to be in-depth researched. At present, the existing literatures lack the research on attack survivability of weighted network under imcompleted information. Thirdly, most of the studies about survivability are based on nodes' failures, but the real situations are that most failures are edge failure, thus, the survivability of edge needs to be further investigated in weighted networks. Finally, most of studies in complex networks are limited on the topology factors, while, using recovery/protection technologies are able to more effectively impove the survivability of weighted complex networks. Based on above matters, the main work of this paper as the following:1. This thesis analyzed the weighted complex network evalution models and their metrics for measuring the survivability. According to the characteristics of the weighted complex networks, this paper analyzed the weighted network survivability based on the Betweenness in incomplete information scenes. We defined the term of importance degree of nodes, and used this defination to describe the importance of different nodes in networks. We focused on the following aspects:1) how to obtain nodes'information,2) attack strategies,3) the proportion of attacked nodes. The simulation results indicate that hiding just a small fraction of vertices randomly can enhance the invulnerability and obtaining just a small fraction of important vertices can quickly destroy a network. This conclusion can guide people based on the importance of nodes to imply corresponding recovery or protection technologies.2. This thesis proposed a survivability model named F(a,I,f) based on the edge faults. By analyzing the robustness and efficiency of weighted complex network in random attack, incomplete information attack and intentional attack scenarios, we examined weighted complex network survivability. Through the simulation results, the author got a conclusion that the weighted network has a strong invulnerability to random attacks, while it is very fragile under intentional attack.3. By studying the characteristics of IP layer and existing IP layer network fault recovery technologies, in order to enhance failure resiliency without jeopardizing routing stability, this paper proposed an Imperative-Forwarding-Path-Routing (IFPR) scheme. The proposed sheme can offer full repair coverage for signal link failure. The simulation results demonstrated that IFPR can enhance networks' performance and has better stability... |
PDF Full Text Request