一种基于完整性检测的Agent迁移容错模型  

A NOVEL AGENT MIGRATION FAULT-TOLERANCE MODEL BASED ON INTEGRITY VERIFICATION

作　　者：蒋嶷川[1] 钟亦平[1] 张世永[1] 

机构地区：[1]复旦大学计算机与信息技术系,上海200433

出　　处：《模式识别与人工智能》2004年第3期267-274,共8页Pattern Recognition and Artificial Intelligence

摘　　要：目前Agent的容错研究中,一般都是通过在Agent的每一步迁移过程中产生复本(Repli(?)ation),然后通过选举(Voting)得到结果.但是大量Agent复本会浪费网络资源与时间.为此,本文提出了一种基于完整性检测的Agent迁移容错模型NAMFTM及其改进模型P-NAMFTM.该模型能有效降低通信复杂度与复本数量,避免大量冗余Agent复本所带来的资源负载与时间耗费.最后,本文构造了P-NAMFTM的π演算模型,采用π演算对其进行了分析验证,演算结果证实本文提出的模型是正确可行的.Of other researches of agent fault-tolerance, the method of replication and major voting is often adopted. However, with such method, many agent replications are produced in every agent migration step, which may cost much network and time resource. Aiming at such situation, this paper provides a novel agent migration fault-tolerance model(NAMFTM)and its improved version(P-NAMFTM). The model avoids the cost of network resource and time brought by many agent replications. At last, the paper constructs the π —calculus model for P-NAMFTM, and then makes analysis for it based on π-calculus. The π-calculus analysis result proves that the model provided by the paper is correct and valid.

关 键 词：移动AGENT 恶意主机问题 Agent容错 Π演算 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]