Research On Dynamic Real-Valued Clonal Selection Algorithm

With the rapid development of network technology, computer network has penetrated into all aspects of social life, and network security problems are more and more arousing people's concern. Compared with the static firewall technology based on data analysis, dynamic intrusion detection has the unparalleled advantages. Intrusion detection system based on immune principle has become a new research hotspot after neural networks and evolutionary computation. Dynamic clonal selection algorithm, as a core algorithm of intrusion detection system based on immune principle, is of high research value.Binary encoding method is not easy to search in large space and has many other disadvantages, while the real-valued encoding method overcomes these shortcomings, so in this paper, the dynamic clonal selection algorithm study and simulation experiments are done based on real-valued encoding method.In this paper, by studying a variety of information processing mechanisms that the immune system contains, as well as the latest computer immune theoretical results, the following two improvements of the existing dynamic clonal selection algorithm has been carried out:First of all, we makes the system to avoid repetitive negative selections on memory detectors, matured detectors and a part of immature detectors, by restructuring of the recent dynamic clonal selection algorithm, to reduce the algorithm's time complexity.Secondly, we present the use of a balanced binary tree structure to overcome the duplication of memory detectors to avoid duplication of alarm.At last, we made a simulation, using the KDD CUP 1999 offline intrusion detection data. The results show that the improved dynamic clonal selection algorithm has been improved significantly in time complexity and memory detector efficiency in the generation of performance.