基于免疫机理与合同网协议的多Agent入侵检测系统  

An immune and contract net protocol-based multi-agent intrusion detection system

作　　者：马鑫[1,2] 梁艳春[1,2] 田野[1,2] 于涛[3] 

机构地区：[1]吉林大学计算机科学与技术学院,长春130012 [2]吉林大学符号计算与知识工程教育部重点实验室,长春130012 [3]中国科学院长春光学精密机械与物理研究所,长春130033

出　　处：《吉林大学学报（工学版）》2011年第1期176-181,共6页Journal of Jilin University:Engineering and Technology Edition

基　　金：'863'国家高技术研究发展计划项目(2009AA02Z307);国家自然科学基金项目(10872077);吉林省科技发展计划项目(20080708)

摘　　要：建立了一种基于免疫机理与合同网协议的多Agent入侵检测系统ICNPIDS。在被动免疫抗体PAb、记忆自动免疫抗体MANAb及模糊自动免疫抗体FANAb的基础上,将合同网的协同方法应用到抗体检测中,提出了联合免疫抗体UAb的概念,并研究了UAb抗体的生成、检测和更新过程,从构造上克服了Agent间分析经验难以共享借鉴的问题。通过滥用检测和异常检测技术的使用,显著提高了系统的检测性能,增强了系统对于外界负载变换的适应能力。模拟实验结果验证了ICNPIDS是一种具有较高检测性能、可自主适应环境变化的入侵检测系统。A multi-agent intrusion contract net protocol. Applying detection, the concept of united immune antibodies, the memory generation, detection, and updati detection system （ICNPIDS） was proposed based on immune and the cooperation method of the contract net protocol to intrusion immune antibodies was also proposed by analogy with the passive automatic antibodies, and fuzzy automatic immune antibodies. The ng of the united immune antibodies were examined, which overcomes the difficulty that the analytic experience can not be shared between agents. The efficiency of the intrusion detection model that uses abusive detection and the analogy detection was improved greatly. Experiment results show that ICNPIDS is a more adaptive and efficient system.

关 键 词：计算机应用 网络安全 入侵检测 人工免疫 合同网协议 联合免疫抗体 AGENT 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]