The Principles Of Signals Presenting And Danger Sensing In Computer Immune Systems

Computer Immune Systems, Which are also called Artificial Immune Systems, are biomimetic systems.These approachs simulates the principles of Biological Immune Systems, and they are adaptive defence systems.The Biological Immune Systems are made up of two layers, one is Innate Immune System and the other is Adaptive Immune System, the researches of Computer Immune Systems are also related to the two layers.Negative Selection Model is the most popular model in Computer Immune Systems, the biological Theoretical basis of which comes from adaptive immune system. The essential of Negative Selection Model is to train lymphocytes with Self, delete those which can match Self and leave the ones which can detect Nonself. Negative Selection Model has the ability of detecting unknown Nonself, with features of adaptability and diversity, and these are consistent with the intelligent features of Biological Immune Systems. However, both the Self and Nonself sets are infinite and uncertainty, which cause the High computational complexity and high false recognition rate.Danger Theory is a new hypothesis in innate immune system, the main viewpoint of Danger Theory is that the function of immune system is to find out potential danger and keep the balance of a system. So the immune system is not required to distinguish huge amounts of Selves and Nonselves, it is only necessary to concerned about whether the system is in danger or not. AS a complementarity of Negative Selection Model, introducing Danger Theory to Computer Immune Systems, can avoid the massive Self/Nonself processing problem.As a new Model in Computer Immune Systems, the research of the Danger Theory is immature, The key points such as the presentation of danger signals, the determination of danger states are still reliance on human experience, which makes the current danger Theory immune model less of adaption and diversity and losed the intelligence which a Computer Immune System should have.In order to build an adaptive Computer Immune system based on Danger Theory, and reflect the intelligence of immune models, this paper took the danger signals and artificial antigen presenting cells as main research objects.A realizable scheme of a series of problems, such as definition, presentation and fusion of danger signals, apperceive of danger states were presented.Differential is a conception in mathematics, which is used to describe the changes of a function.In this paper, Differential was also used to describe the changes of the computer systems and the changes of computer systems were defined as danger signals.For the data in a computer system are discrete, Numerical Differentiation was taken as the theoretic basis. More over, several feasible calculating method of danger signals were presented in this paper.Refer to the function and principle of the antigen presenting cells, artificial presenting cells were build in this paper. The function of artificial presenting cells were detecting and fusing danger signals, abstracting macroscopical system states from microcosimic danger signals and realizing danger apperceive. The structure, and lifecycle of artificial antigen presenting cells, especially the function, structure and relative algorithms of the Toll like Recepters which are the key part of artificial antigen presenting cells were discussed in this paper.Finally, a bot and a worm with typical latent characteristics were taken as examples to validate the feasibility of the danger theory model presented in this paper. The experiment include four group data, with the control experiments, the latency softwares could be detected with the danger signals and artificial presenting cells, and the danger model was proved to be feasibility. With the analysis of compositions of Toll like Recepters in different phases of the latency softwares, we found that the compositions of the Toll like Recepters and the categoris of danger signals change with the run phases of latency softwares, which proved that the danger model in this paper is adaptive. Another group of experiments showed that the compositions of Toll like Recepters were different in bot and worm, which proved that the danger model in this paper had the characteristic of diversity.