Study On Approach To Complete Attack Graph Auto-generation And Analysis

While bringing us the convenience, the rapid development of the Internet also brings a lotof security issues, which are caused by the vulnerabilities of the hosts in the network. Duringthe attack, an attacker often uses multiple vulnerabilities to achieve his purpose, thusdisplaying and analyzing the relationship between the host vulnerabilities means a lot for us.The attack graph is just such a tool.Firstly, this paper gives a new method that obtaining the network connectivityautomatically by analyzing the firewall configuration files, to get rid of tedious manual input.A model based on the method is built and tested. The test result shows that the running time ofthe model is substantially independent of the network scale, and the method can be applied tolarge-scale network.Secondly, as the generation of attack graph without loops leads to missing of attack paths,this paper puts forward the concept of complete attack graph. Then the attack patterns areenriched to cover almost all network attack types and based on them an efficient approach tocomplete attack graph generation is given, whose complexity is between O(N2) and O(N3). Amodel to generate complete attack graph automatically using the algorism is built, which isshowed practical and efficient in the test.Thirdly, the particle swarm optimization algorithm is introduced into the optimizationsecurity measures analysis for the first time in this paper. The paper gives an optimizationsecurity measures analysis algorithm based on the binary version of PSO, and builds a modelusing the algorithm. In order to use the model to analyze the complete attack graph, the paper proposes an approach to get sub-graph without circle from complete attack graph and ascheme to quantize graph node repairing cost based on the CVSS scoring system. The testresult shows that the optimization security measures analysis algorithm proposed by thispaper can provide users with a variety of meaningful solutions, and it is more practical andaccurate than the other optimization security measures analysis algorithms.