【作者】 李少玲；

【导师】 韩芳溪；

【作者基本信息】 山东大学 ， 计算机应用技术， 2007， 硕士

【摘要】 迁移工作流(Migrating Wlorkflow)是将移动计算技术引入传统的工作流管理系统而得到的一种新技术，是近年来工作流管理研究的一个新方向。迁移工作流管理引擎、迁移实例和工作位置是迁移工作流管理系统的三个要素，工作流引擎完成工作流过程定义、迁移实例生成和多迁移实例协调等工作；工作位置由停靠站服务器及其关联的工作机网络组成，是迁移实例的运行场所；迁移实例是以移动Agent为范型构建的，由迁移工作流引擎或工作位置创建，是任务的执行主体，可以在某个工作位置上利用本地资源执行一项或多项任务。多个迁移实例通过迁移和协作完成工作流过程。迁移工作流系统中可靠的组播通信机制是迁移实例之间协同工作的一个关键保障，也是迁移工作流系统研究中的一个重要问题。论文借鉴支持分布式协同工作的组播通信及移动Agent组播通信的研究成果，结合迁移工作流系统的特点，提出了一种基于逻辑令牌环结构的迁移工作流实例组播通信模型，并根据该模型设计了保证实现可靠组播通信语义的全序组播通信协议。文章中详细的给出了该协议的数据结构及相应的组播通信算法，并对协议的正确性进行了证明，对协议的性能进行了分析。该协议使用令牌来实现组播消息的次序化，保证了组播消息的全序性；通过设计适当的缓冲机制来保证消息的可靠传输及组播消息传输的原子性，有效的解决了组播通信过程可能存在的全局不一致的问题，确保了迁移实例可靠的组播通信。本文根据该组播通信模型，设计了一种容错的基于层次结构的组成员管理机制，该机制能够有效的实现对组成员的加入、退出及出错的管理。层次结构的设计有利于组播通信系统的扩展，使得该系统具有较强的伸缩性。针对组播通信过程中可能存在的成员失效问题，提出了一种随机心跳故障检测方法，能够及时检测发生错误的组成员并将它移出通信组，实现了可靠组播通信系统的差错控制。论文首先对工作流、迁移工作流和可靠的组播通信基础知识进行概述；而后根据迁移工作流系统模型，提出了基于令牌的迁移工作流组播通信系统模型，并根据该模型设计了全序的组播通信协议，给出协议的数据结构及相应的组播通信算法步骤；随后提出了基于层次结构的容错组成员管理结构及基于工作位置的随机心跳故障检测方法，并对该方法的正确性进行了分析。更多还原

【Abstract】 The migrating workflow is a new technology applying mobile computing to workflow management, which is a new direction within the research of workflow management. Migrating workflow management engine, migrating instance and work station are three main elements of migrating workflow management system.Workflow engine complete the definition of workflow process, the generation of migrating instance and the coordination of more migrating instances.Work station refers to anchorage and network associated with it,wich is the operation place of migrating instance. Migrating instance is constructed by the mode of mobile agent, created by workflow engine or work station, is the main implementator of task, to complete one or more tasks by using the local resources of work station. Workflow process is completed by a number of migrating instances’ migration and collaboration. Reliable Multicast System is a key protection mechanism to realize the cooperation of migrating instance, which is also an important issue of the migrating workflow research.From the research results of distributed cooperation and mobile agent multicast, based on the characteristics of migration flow, a multicast system model based on token is proposed. A multicast protocol is designed to ensure the realization of the entire sequence of semantic reliable multicast according to the model. The data structure of the protocol and the corresponding algorithm are given by this paper in detail.The correctness of the protocol is also proved in this paper. The protocol can resolute the the overall inconsistency problems of the multicast effectivly. The entire message sequence of the protocol is realized by the use of the token, ensure the reliable transmission by the appropriate design of buffer mechanisms, effectivly solute the overall inconsistency of mulicast, and ensure a consistent state of mukticast. According to the multicast model, a fault-tolerant group member management mechanism based on hierarchical structure is designed in this paper.The agreement to join the group or from the group or the fault management is effectivly realized in the mechanism. This hierarchical structure is also conducive to expansion of the multicast system, and also has highly flexible.For the possible failure of members in multicast, a random heart failure detection method is proposed, which can detect the error members in time and realize the error control of the multicast system.First, the basis knowledge of the migrating workflow and multicast is outlined in this paper. Latter, according to the migrating workflow mode, a mulicast system of migrating workflow based on token is designed, the whole sequence of multicast protocol is also designed, the data structure of the protocol and the corresponding algorithm are gived too. Then, a fault-tolerant member management structure based on the hierarchical structure is proposed and a random heartbeat fault detection method based on the location of the work is designed, the method’s correctness is also analysised in this paper.更多还原