Research On Immune Intrusion Detection Technology Based On Dendritic Cell Algorithm

In this recent world,more and more security problems in the internethappen,as long as the technology of internet grows. The traditional intrusiondetection such as firewall,is able to defend some simple intrusions while is notable to fight against active intrusion attack and it is very difficult to cope with allkinds of complicated network attack. The features self-adaptive、self-organizedand distributivity owned by immune system are the very characteristics that isexpected by the traditional intrusion detection technology. Therefore,more andmore experts have focused on the intrusion detection technology based onartificial immunity.Artificial immune system is inspired by biological immune system,then doanalog and simulation,and it has been applied on lots of aspects. There are someproblems that can not be explained in the intrusion detection system based ontraditional artificial immunity theory,while some newly Immunity theory——danger theory have been proposed along with further study on Biology. Andnow,Innate immunity and self-adapt immunity have both been drawed intoArtificial immune system. The research indicates that it is vital of Dendritic Cellin the immunity system,thus Dendritic Cell Algorithm has been proposed resultedfrom dendritic cells. This article did some other study on the basis of theDendritic Cell Algorithm.At the first,this article has done some study deeply on Dendritic CellAlgorithm. There are some problem of classification for antigens in thetraditional Dendritic Cell Algorithm,resulting high false alarm rate,whileDempster/Shafer theory is the very algorithm to solve the problem that DendriticCell Algorithm can not cope with the uncertain,thus the Dendritic Cell Algorithmbased on Dempster/Shafer theory is proposed. We add Dempster/Shafer theory tothe traditional algorithm. And the proposed algorithm solved the problem of high false alarm rate,and could also improve the ability of detection.Then,a new intrusion detection model was proposed. In this model,we didsome improvement based on Dendritic Cell Algorithm,and make sure about whatthe purpose of each step so that the partial of the system is the best. We haveanalyzed the fundamentals of intrusion detection in the network, illustratingevery component of the model. At the last,we test the new model with DCA、SVM and the improved algorithm using kddcup99data set,and the resultindicates the model succeeds in lowering false alarm rate,besides,the detectionrate has been improved.