Research On Evaluating Method Of Network Attack Effects

Research of cyberspace has become one of the hot. The network war will become the core of thefuture Cyber-War. In Network War, how to evaluate effects of network attacks is an important andchallenging subject in cyberspace theory.The main idea of paper evaluates attack effects based on “attribute-atomic oparation-atomicindex” evaluateon indices system, by means of change of evaluation on atomic index in target systembetween before attacks and after attacks. With the idea for guide, model and methods of networkattack effect evaluation and realization scheme are developed thoroughly, main works as follows.1) As the correct understanding of network attack is a prerequisite, we propose acomprehensive-attribute evaluation atomic model based on the accurate description of network attack.The model is divided into three layers. The first layer is the attribute layer, the second layer is theatomic level and the thrid layer is the index level.The attribute layer, which is a man-made description of comprehensive-attributes when assessed,is relatively rough.and it is easy for experts to score by AHP. The implementation of the minorityfollows the majority solves the disagreement, which is inconsistent in the scoring of the experts. Theevaluation is optimized by the introduction of screening factor and compensated factor.The atomic layer as an intermediate layer, one hand the network attacks can be described indetail; on the other hand, atomic index can be defined in the corresponding target system. The atomicoperations are obtained by mining the network attack database.The weights of the atomic indexs in the index layer are calculated by introducing the concept ofrough set.As they need data not any priori knowledges, the weights are more objective and eliminatesubjective factors. According to the atomic indexs are too much, the atomic indexes are screenedsecondary by Delphi Method.2) Takeing SYN Flood attack as an example, we calculate weight coefficients of each layerintegratly.Then, we gained the threat coefficient after introducing the evaluation matrix and the datacollected through the normalized evaluation. Finally, combining data provided by the U.S. thinktanks with the characteristics of the CAA model, we analyse the possible errors.