Research Of Detector Generation Strategy In Immune Intrusion Detection

With the continuous development of the modern Internet,network is critical of human life and it has been change it, the development of agriculture and industry are based on Internet, the promotion of knowledge and the exchange of information also changing greatly. Network make our life better and easier, but it also brought us a lot of security problems, the system is malicious intrusion, theft of personal information, security issues become more and more important.Immune intrusion detection as an active safety mean also becomes a hot research.Immune intrusion detection technology is a proactive network security defense of artificial immune way through autologous tolerated,it can construct an effective detector in response to changing attack. Detector is core of intrusion detection system, which uses negative selection algorithm to generate the mature detector and then to match the attack. In process of this, mature detector constantly matched with itself to achieve dynamic updates. So intrusion detection system have a higher detection rate and robust, and because the system has a good initiative and self-learning ability, so that the immune intrusion detection system is an effective way of defense.To deal with the problem of the slow data processing speed of immune intrusion detection and poor real-time detection, Nonnegative matrix factorization by Bregman iteration is proposed which improved the traditional method,changes matrix iterative process,using the matrix location to realize the decomposition conditions and its constraint, better retention of the internal structure of the data and acceleration of the processing.At the same time, through improved co-evolution approach, using a varietyof main-center individuals and quantum variation mode,increased the cooperation way of detection process and detection process, To speed up the information exchange of detection, and make detector system's update iteration more effectively.improved evolutionary algorithm's weakness(EA) of slow convergence and prematurity, effectively improve the detection rate and detection efficiency.Experiments in KDD CUP 1999 data sets show that the approach can improve the speed of intrusion detection and enhance the timeliness of immune intrusion detection.