Vulnerability Detection Techniques Based On Research Of Complex Network Topologies

Network, such as WWW network, protein interaction network, power network, interpersonal network and biological chain network that in real life networks are closely related to our daily lives. How to design and manage these complex networks has a big influence on the nature of complex network, topological structure, and robustness.First, it gives a brief introduction of related conception of the basic theory of complex networks and its vulnerability in Chapter Two. Then, it defines the complex network topology Vulnerabilities from the node level to the structure of the network. Thus, a vulnerability analyzing method based on critical nodes and critical cluster is proposed.Second, concerning the analysis of critical nodes, these papers deeply study the realization method and its advantages and disadvantages of current classical detection algorithms of critical node. It gives a full consideration of the actual network nodes characteristic has weight value. Therefore, based on Page Rank algorithm, Page Rank key node detection algorithm is proposed based on weight and it introduces decomposition transfer matrix method that can reduce unnecessary calculation and lower the complexity of time and space, it improves the precision of calculation and effectiveness as well.Third, on the analysis of critical cluster, this paper links it with complex network community structure and deeply studies the realization method. The advantages and disadvantages of classical community structure partition algorithm are studied in this paper as well. In order to study it, label propagation algorithm is introduced in this paper. The label propagation algorithm is put forward on the basis of module degree to solving problem such as random updating, unstable results, small community and other problems that exist in the process of algorithm transmission. Binding the close nodes before propagation can reduce unnecessary iterations. It chooses label according to the sequence of node degree in the process of algorithm and use greedy algorithm. Whether to update the label or not is according to a single community modularity increment, it can avoid the label “counter flow” phenomenon. The algorithm does not change linear time complexity and avoids the generation of small communities, so it improves the quality of community division and helps us to find hidden communities.Finally, this algorithm is verified to be correct in the tests. The comparison of test results is displayed in simulation mode.From the perspective of complex network topologies structure, this paper does a research of the complex network nodes and cluster's vulnerability. The research on complex network topology's vulnerability has great significance for network structure's optimization, epidemic prevention and virus attack's preparedness, etc.