Research On Dynamic Adaptive Mechanism Of Self And Detector In Immunity-based Intrusion Detection System

With the information industry developments, there are more and moreproblems of network and information security. As a network security technologywhich has the initiative character, the research of intrusion detection system isattacting more and more sights of researchers. Moreover, the advantages ofbiology immune system, such as, self-adaption, self-organziation, dynamics, andso on, are applicable to the intrusion detection system. Sothat, immunity-basedintrusion detection system has being became a research hotspot. The central parts:self and detector set play a major role in detection performance. For that reason,this paper focuses on self set and detector set and their update, optimizationmethods, learning mechanism, etc. in immunity-based intrusion detection system.The self set in the immune-based intrusion detection system which is used totrain detectors has some defects, especially, in the real-valued shape space: multi-area, overlapping, noising sample, etc. which can cause some problems, such asthe boundary holes of detector set, the high cost of generating detectors, etc. Tosolve the problems, a self set optimization algorithm is proposed, which usesfuzzy clustering algorithm and Gaussian-distribution theory. The fuzzy clusteringdeals with multi-areas and the Gaussian-distribution deals with the overlappingand noising. Experimental results show that, the optimization algorithm can solvethe selves’ problems, increase the efficiency of detector generation effectively.The self set is made up of the empirical data which can not mirror thecurrent real facts with the changing environments. This would lead to problemswith the detector generation. To solve the problems, borrowing ideas fromsupervisory mechanisms, immune feedback theory, etc. in biology immue system,a self set real-time update algorithm is proposed. This algorithm includes some modules and auxiliary sets which can deal with the different operations andupdate the self set collaboratively and let the selves keep pace with the changingenvironments. Experimental results show that, the algorithm can achieve theintended purpose, lay solid foundations for detector generation.Detector set, the most important role in the immunity-based intrusiondetection system, also has some problems, especially, in the the real-valued shapespace. The holes and overlapping problems has not been solved effectively before.To solve these problems, an optimization algorithm for detectors is proposed,inspired by the immunoregulation of immune cells in biology immune system:Updating the detector set by the candidates generated from their parents and theaffinity comparison to improve detectors’ distribution performance. Theexperimental results show that the optimized detectors can increase the efficiencyof detectors’ distribution and improve the detection performance of detector set.The detectors’ dynamic update impacts on the detection performances. Inthe biological immune system, the dynamics of self cells and immune cells is toensure the system adapt to the changing environments. And in the biomedicalengineering, the vaccine mechanism is an important role on fighting off the virusviolations. To deal with the detector problems, this paper proposes an adaptivelearning algorithm of detector set with standpoint of the vaccines and dynamicdetectors. It contains some detector learning modules and keeps the detector setlearning and adaptive adjusting with environment. Experiments results show that,the algorithm can let the detector set keep a real-time understanding on thechanging environment and keep the detection performances.The researches on the self set and detector set which are the most importantroles in immunity-based intrusion detection system analyze the problems, mainlyfocus on the optimization algorithms, the dynamic update and adaptive learningalgorithms to increase the system’s detection performances. The contents areuseful for not only the immunity-based intrusion detection system, but also theartificial immune theory and its correlative research fields.