The Research Of Immune Negative Selection Algorithm And Its Application In The Worm Detection

According to the developed basic principles of biological immune system, theartificial immune system is mainly used in all types of information processingtechnology, computing and intelligence systems. The negative selection algorithm ofthe artificial immune can be described as the problem that can recognizes "self" and"nonself"and eliminate the "nonself". Learning from the immune negative selectionmechanism improves computer security, namely, the simulation research on naturalimmune system plays an important and practical role in researching at computersecurity system.Consideration at shortcomings of existing negative selection algorithm, realnegative selection algorithm is improved and two improved negative selectionalgorithms are proposed in this thesis. First, a two-dimensional spatial radius variablenegative selection algorithm is proposed aiming at real number coding negativeselection algorithm in order to improve the algorithm detection efficiency. In addition,coding negative selection algorithm is improved based on binary and a kind of rvariable and haiming rules adaptive fusion of negative selection algorithm ispresented. The use of nonlinear chaotic generating detectors can assure the detector’sdiversity. And a kind of adaptive fusion operators, which realize r variable andhaiming rules adaptive fusion, is designed. In this thesis, programming helps toimprove the simulation algorithm, and experimental simulation result indicates thatblack hole amount is greatly decreased and black hole space is reduced whencomparing the improved algorithm simulation data with the original one. Meanwhileblack hole space detection rates and detection cover space have been greatly improved,which proved the improved algorithm effective.In the end of the thesis, two-dimensional spatial radius variable negativeselection algorithms are applied to computer worm detection, and a kind of improvedworm virus detection negative selection algorithm model is designed that make theuse of variable radius solid value to produce negative selection algorithm with asmuch coverage as it can make to detect the worm attacks. Experiments show thatworm detection algorithm has the characteristics of accuracy and real-time.Experimental results indicate that the proposed method yields high detection rates forobfuscated viruses with an averaged recongnition rate of90%in real world conditions, the false positive rate can be maintained below5%.The method has a goodgeneralization ability.