Research On The Applications Of Artificial Immune In Network Intrusion Detection System

With the development of Internet, It is more convenient to use various tools to attack network. And these cause great financial losses to society. So people pay more attention to the network security. Traditional technology of network security can not fit the more and more complex network environment. The good performance of human immune system protect body against the invasion of pathogens inspire people to establish the artificial immune system. The similarity of function of intrusion detection system and the human immune system make people widely use artificial immune system in the network intrusion detection system. And in this way, the function of intrusion detection is promoted.In this paper, we introduce the development and status of network intrusion detection based on artificial immune system, the basic knowledge of intrusion detection, the concepts and mechanisms of human immune system, the main algorithm and development of artificial immune system. By analyzing the dynamic clonal selection algorithm, we proposed the method of improvement:First, the dynamic clonal selection algorithm generated immature detectors by gene library evolution. But when the gene segments become huge in the gene library, they will cluster. This phenomenon will reduce the diversity of detections, at last influence the result of the intrusion detection system. So we propose optimizing the gene library by cluster algorithm. And using the operator of deleting and mutating to change the segments of gene in the gene library.Secondly, in the dynamic clonal selection algorithm, the population of memory detections is fixed. When the attack frequently in the network is changed, the memory detections can not work better. We propose that dynamic changing the population of memory detections by the changing of attack frequently and hope to enhance the adaptability of the intrusion detection system.At last, We establish a model of network intrusion detection system based on improved dynamic clonal selection algorithm. By simulation experimenting, it is proved that the improved algorithm enhances the detection results and the adaptability of the system.