The Research Of A Danger-inspired Artificial Immune Model Based On Fuzzy-set

Currently,almost all of the immune models in intrusion detection are based on the traditional"self"and"nonself"(SNS) theory. In this theory, the organism immune system doesn't produce the immune response to"self"antigen, but will trigger the immune response to the"nonself"antigen and eliminate it. In order to overcome the shortage of SNS model, some scholars advanced danger theory (DT), which is new, and the"self"and"nonself"also needed to be distinguished in it, but the danger signal is the key factor producing the immune response not the"nonself"required in the SNS model.On the foundation of have analyzed the existing immune-inspired danger theory models, combining with the concept of fuzzy set and in order to respond the questions such as the uncertainty of the risk domain and the imperfect of the abnormal detection system, a new immune danger theory model based on fuzzy set was researched in this paper. The date stream was disposed with three levels in the new model, the first abnormal disposal module carried on detecting the danger and isolating the no hazardous as well as non-invasive date, the object disposed in the second module was dangerous and abnormal incursive data, which was carried on an appropriate danger classification, and finally the danger would be eliminated in the third level module. Besides, for the problem that it's difficult to determine the risk region, there were two methods used to define it in this paper, which were the danger signal definition based on genetic optimization (DSDBGO) and the danger signal definition based on fuzzy-set (DSDBF). The former was to define the danger signals on the basis of genetic optimization and self-adaptive feature subset selection of the immune genetic algorithm, while the latter carried on defining the danger signals on the basis of fuzzy set as well as introducing the concept of membership in fuzzy set and subordinate degree.The immune danger theory model based on fuzzy set in this paper was researched on the foundation of the danger signal definition based of fuzzy-set (DSDBF), the danger detection algorithm (DDA) and the danger elimination algorithm (DEA) were also studied and designed in it. As it's showed in the experiment that the danger detection algorithm (DDA) designed in this paper was better than the abnormal detection algorithms in other intrusion detection models, because it improved the recognition efficiency for organism to"danger", advanced the accurate rate efficiently of immune-inspired intrusion detection system as well as reduced the false alarm rate.