Research On Real-value Negative Selection Algorithm And Antibody Optimization In WSN IDS

Wireless sensor networks with its unique characteristics of distribution, self-organization, adaptive, miniaturization and wireless connection, are widely used in industrial production, environmental monitoring, health care, transportation and other fields. Now, with the development of wireless network and the popularity of intelligent terminals, wireless sensor networks are more and more promising, which is confirming the prophecy "one of the most influential technology in twenty-first Century". Same as the traditional network, it is faced with a variety of intrusion too, besides, the deployment environment is changeable, energy and computing power is limited, so the wireless sensor network security field has some special invasion, as a result, the traditional intrusion detection system can't fully adapt to the wireless sensor network. The self-organization, dynamic adaptation, robustness, distributed, and self-learning of the artificial immune system fit the wireless sensor networks so exactly that wireless sensor network intrusion detection system based on artificial immune system come into being. It is the reason of IDS in WSN based on immune having innate advantage, making it become a new focus.Through the analysis and summary of existing intrusion detection system based on immune, aiming at the problems that the detection result of existing system is not satisfactory and the system can't effectively deal with large-scale intrusion when exiting system wildly coding with binary, this paper presents a kind of algorithm named RNS-WSN adapting to wireless sensor network, in which antigen / antibody is coded with real valued, and simplifies gene to reduce the occupied space and the operation process, replacing r-continue method with Manhattan distance to compute affinity, moreover, according to the characteristics of intrusion to reduce the randomness gene and improve the affinity of antibodies, the fact verified by experiments on platform NS3 is that applying RNS-WSN algorithm can obtain better detection results than traditional binary coding, can effectively deal with large-scale invasion with same energy consumption.For the problems that the deficiency of the affinity and poor pertinence of antibody, to reflect the dynamic adaptation of the system, inspired by the idea that picking gene cross layer and mixed can improve diversity, this paper proposes a scheme: according to the commonness of intrusion in species layer, when facing different intrusion, the ratio of gene from this layer should reasonably expand to improve the affinity and to produce antibodies having stronger pertinence, that can lead better detection effect. Similarly, the scheme is verified by experiments. The results show that extracting gene according to network layer intrusion occurred and mixing gene proportion properly can lead higher binding between antibody and antigen, making antibody more targeted when facing different intrusion.