Intelligent Detector And Evolutionary Method For Network Layer Of IoT Based On Immune Danger Theory

With rapid development of computer and network technology, more and morepotential safety problems are exposed. Inspired from biological immune system,People hope to build a defense system as powerful as biological immune system toprotect computer and network. With high false alarm rate, high non-response ratesand low efficiency, traditional artificial immune system hard to adapt to the complexnetwork. Firstly, immune danger theory makes up for the defects of self-non selfselected on traditional artificial immune system, reduce computational cost andimprove efficiency. Secondly, the Internet of things is multidimensional,heterogeneous, dynamic network which have three layers. It is similar to biologicalimmune system. Therefore, it is so important to research detector and evolutionarymethod on the network layer of Internet of things. And it is based on immune dangertheory.At first, this paper analyzes several typical defects of detector generation andevolution method, designs the KTRN detector based on immune danger theory,proposes KRNA detection algorithm. KTRN detector consist unknown dangerdetection module, a known risk detection module, the adaptive module, memorymodule, intelligent module, the module and test auxiliary module. KRNA detectionalgorithm match abnormal data and danger unit in memory module by r continuousbits matching method. And then calculate danger rate of abnormal data by collectingkey value change method and detect data is dangerous or not. It shows that the KRNAdetection algorithm has higher detection rate, lower false detection rate and reducesthe redundancy, eliminates black hole in experiment. secondly, DSAA dynamicadaptive algorithm is proposed, it based on the problem of traditional detectionalgorithm difficult to adapt to environment.The algorithm get occurrences of all kindsof danger data in a day, and get a new danger frequency by averaging result of addingit and danger frequency of such danger data, and adjust the number of organelles by the new danger frequency. It shows that the algorithm improve adaptive ability ofdetector, reduce the detection response time in experiment. At the last, PTRA detectorpopulation evolutionary algorithm is proposed based on problem of the traditionaldetector evolutionary algorithm population diversity is poor, adaptability is weak. Itshows that the algorithm is better in diversity of population in experiment. And it alsoimproves individual fitness obviously on the basis of keeping detector ability inexperiment.