Based On Idiotypic Artificial Immune Network Model And Application

In recent years, biological immune system-based artificial immune systems have become a new direction for computational intelligence, and applied to optimal control, pattern recognition and computer security and other fields. In the intrusion detection technology, especially in the denial of service attack detection, the immune system was able to play to its unique characteristics. By idiotypic immune network theory of reference, we can very well against denial of service attack on the conduct of identification.The thesis introduces the biological immune system and the traditional characteristics of artificial immune algorithm first, and then gives the detailed explanation of intrusion detection field. At the same time focusing on the idiotypic immune network and its mathematical model is introduced. On this basis, it puts forward a new idiotypic immune network model (IINM) and applied it to the denial of service attack detection.The traditional denial of service attack detection is difficult to identify with the characteristics, even if there are significant errors identified, such as the concentration of the normal behavior of network traffic flow identified as abnormal. Meanwhile, in certain rare types of denial of service attack simply is not recognized. By applied IINM model, due to the idiotypic immune network's own immune characteristics of the aggressive behavior of the cluster becomes very easy. Thus, by importing test data set, we can visually see gathering area of attacks.For a rare type of denial of service attack detection, we adopted the changes in coefficients of the coordinate axes, reduced the phenomenon of a large number of common type of clustering reaction cover up the key features of the rare type.In the end of the thesis, we generalize the situation of denial of service attack detection, and evaluate the accuracy and effectiveness of the model from experiments. The result shows it is better than the traditional denial of service attack detection method.