摘要： 针对传统入侵检测技术难以适应动态的网格计算环境等问题，依据免疫原理，提出了一种基于Multi-agent的网格入侵检测模型(GIDIA)。描述了GIDIA的体系架构，给出了免疫模型、检测Agent、决策Agent和防御Agent的定义，建立了相应的抽象数学模型及推理方程。理论分析和仿真结果表明，GIDIA解决了信任社区内与社区间的协同预警及防御问题，具有检测率高、自适应能力强等特点，为实现网格安全提供了一种新方法。

关键词: 网格安全, 入侵检测, 免疫性, Agent

Abstract: Being that conventional intrusion detection systems can not adapt to the dynamic grid environment, grid intrusion detection model (GIDIA) based on application of immunity and multi-agent is proposed. In succession to describe the architecture, definitions of immune model, detective agent, decision-making and preventive agent are given. Relevant abstract mathematical models and detailed inferential equations are founded respectively. Theoretical analysis and experimental results show that GIDIA enables member sites in the same trust community or different ones to forewarn attacks cooperatively, and possesses higher detection rate a with better self-adaptability. GIDIA provides a way for implementation of grid security.

Key words: Grid security, Intrusion detection, Immunity, Agent