WSN入侵检测中实值否定选择算法和抗体优化研究

发布时间：2018-04-01 08:10

本文选题：无线传感器网络　切入点：入侵检测　出处：《哈尔滨理工大学》2016年硕士论文

【摘要】：无线传感器网络以其特有的分布性、自组织、自适应、微型化、无线连接等特点,在工业生产、环境监测、医疗、交通等诸多领域被广泛应用,如今伴随着无线网络发展和智能终端的普及,无线传感器网络越来越被看好,正在印证着“21世纪最有影响力的技术之一”这一预言。同传统网络一样,无线传感器网络也面临着各种入侵,且由于无线传感器网络节点部署环境多变、能量和计算力有限等特点,使得无线传感器网络安全领域存在一些特有的入侵,因而,传统的入侵检测系统不能较好的适应无线传感器网络。人工免疫系统的自组织、动态自适应、健壮性、分布式、自学习这些特点与无线传感器网络恰好吻合,基于人工免疫的无线传感器网络入侵检测系统也就应运而生,正是基于免疫的IDS在WSN入侵检测中存在着先天优势,使其成为新的研究热点。通过对现有的基于免疫的入侵检测系统进行总结和分析,针对现有系统广泛采用二进制编码抗原和抗体导致检测效果不理想,且不能有效应对大规模入侵,以及亲和力计算复杂等一系列问题,本文提出了一种适应于无线传感器网络的RNS-WSN算法,抗原/抗体均采用实值编码,同时简化基因以减少空间占用和运算过程,通过曼哈顿距离替代原先的r-连续位亲和力计算方式,并根据WSN中的入侵特点降低基因的随机性,提高了抗体的可用性,通过在NS3平台上的实验验证,同等能耗下,应用RNS-WSN算法的WSN免疫入侵检测系统能够获得比传统二进制编码更好的检测结果,并可以有效应对大规模入侵。针对现有检测系统中抗体的亲和力不足以及针对性差的问题,为体现免疫系统的动态自适应性和进化性,受基因混合多层次提取提高多样性的启发,本文提出了根据入侵种类的共性,在面对不同入侵时合理扩充对检测率贡献度较大的基因所在层上基因的占比来提高亲和力,生成针对性强的抗体这一方案,来提升检测效果。同样,通过实验对该方案进行了验证,结果表明,根据入侵特点从网络层次混合提取基因,合理配置基因比例,能很好的增强抗原抗体的结合性,使得抗体面对不同入侵时更有针对性。
[Abstract]:Wireless sensor networks are widely used in many fields such as industrial production, environmental monitoring, medical treatment, transportation and so on, because of their unique characteristics of distribution, self-organization, self-adaptation, miniaturization, wireless connection, etc. With the development of wireless networks and the popularity of intelligent terminals, wireless sensor networks are becoming more and more promising, which is supporting the prediction of "one of the most influential technologies in the 21st century." Wireless sensor networks are also facing various intrusions, and due to the changeable deployment environment, limited energy and computing power of wireless sensor networks, there are some unique intrusion in wireless sensor network security field. The traditional intrusion detection system can not adapt to the wireless sensor network. The artificial immune system is self-organizing, dynamic adaptive, robust and distributed. The intrusion detection system of wireless sensor network based on artificial immune emerges as the times require. It is IDS based on immune that has the innate advantage in WSN intrusion detection. Through the summary and analysis of the existing intrusion detection system based on immunity, the widespread use of binary coding antigen and antibody in the existing system leads to unsatisfactory detection effect. Moreover, it can not effectively deal with a series of problems, such as large-scale intrusion, complex affinity calculation and so on. In this paper, a RNS-WSN algorithm suitable for wireless sensor networks is proposed, in which antigen / antibody is encoded by real value. At the same time, the gene is simplified to reduce the space occupation and operation process, the original r-continuous affinity calculation is replaced by the Manhattan distance, and the randomness of the gene is reduced according to the invading characteristics of WSN, and the availability of antibody is improved. Through the experiment on NS3 platform, under the same energy consumption, the WSN immune intrusion detection system using RNS-WSN algorithm can get better detection results than the traditional binary coding. And can effectively deal with large-scale intrusion. In order to reflect the dynamic adaptability and evolution of the immune system, we can solve the problem of insufficient affinity and poor targeting of antibodies in the existing detection system. Inspired by mixed multilevel extraction of genes to improve diversity, this paper proposes a reasonable expansion of the proportion of genes on the layer where the detection rate is large in the face of different intrusions according to the commonness of the invasive species to improve the affinity. In order to improve the detection effect, the method of producing strong specific antibodies is also verified by experiments. The results show that, according to the characteristics of intrusion, the genes are extracted from the network level, and the proportion of genes is allocated reasonably. It can enhance the binding ability of antigen and antibody, and make the antibody face different invasion.
【学位授予单位】：哈尔滨理工大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP212.9;TN915.08

【相似文献】