Research Of Detection Algorithm Based On Clonal Selection And Detectors Distribution

Development of Internet technology is rapidly changing people's living habits.Although it is convenient to our life, it is often faced with many security problems about personal privacy and transaction. Biological immune system is a very efficient adaptive system, and intrusion systems are surprisingly similar with it. Therefore people have been proposed intrusion detection technology based on immune principle. It is a very effective intrusion detection technology it has a very large potential. At present, the immune intrusion detection technology has become a hot topic in the field of network security.This article describes the classification of intrusion detection and related technologies and gives the research status of intrusion detection based on clonal selection home and abroad. Although intrusion detection technology based on clonal selection has made many achievements, there are still many problems. For example,convergence of clonal selection algorithm is too slow, which resulting in a slow response; in the latter, there will be a state of fatigue, decline in the quality of the detector even. To solve these problems, this paper improves clonal selection algorithm with vaccination and Cauchy variation, and the improved algorithm is introduced intrusion detection model. Simulation results show that the improved algorithm can effectively speed up the convergence; in the latter algorithm, you can still improve detection rates and lower false alarm rate.Based on the improved clonal selection algorithm generates detectors population. Although they monomers detection rate is high, between detector duplicate detection problem that the detector uneven distribution of population,covering a variety of data intrusion rate is not high. Now contrast the mainstream intrusion detection algorithm, to learn from their experiences and lessons learned in the population distribution detector problems introduced SVM support vector machine technology. SVM classifiers can efficiently divide the population of thedetector so that the detector can be achieved in a relatively small number of high profile accuracy. The experimental results show that the population has more uniform and higher classification accuracy after SVM classifiers. At the same time, effective classification can greatly increase the utilization of space and resources.