A Novel Negative Selection Algorithm Model And The Realization Of Intrusion Detection

With the continuous development of computer and network popularization, the network security problem increasingly aroused people’s attention. However, the traditional passive defense technology can only detect known attacks due to its own limitations, which can’t meet the demand of today’s growing network. The current intrusion detection system as a new security technology, is a good way to solve the limitation of the traditional passive defense technology. In view of the intrusion detection system has a great similarity with the immune system of human body, which are autonomy, diversity, adaptability, tolerance, distribution and other series characteristics, a new network intrusion detection model is addressed in this paper by leveraging the related theory and mechanism of immune system. The intrusion detection model is using a new type of negative selection algorithm based on deterministic crowding simulating T cells of mature mechanism, which is a kind of artificial immune system based on self/non self-test methodsThe existing negative selection algorithm proposed by Forrest,Perelson et al has many defects which causes the very low efficiency on non-self-detecting. Through experiments implemented in this paper, the reason discovered for this low efficiency is that the algorithm requires a very large amount of detector to completely cover the alien space, however, this is not practical in reality. Therefore, how to achieve the maximum coverage alien space with the minimum number of detector is a urgent issue needed to be solved. Thus, in this paper, a new negative selection algorithm based on improved deterministic crowding is presented. It can solve the problems which the existing traditional negative selection algorithm has in network security, such as the large search space, the low efficiency and so on.The main work in the paper is to improve and enhance the intrusion detection. Proves that the negative selection algorithm based on deterministic crowding can estimate a more accurate method of offset, generate less detector under the premise that merely affects the performance, and also increases the calculation efficiency. The experimental tests show that the proposed negative selection algorithm based on deterministic crowding has higher detection efficiency and accuracy compared with the original algorithm.