Research Of Multi-agent Workflow Access Control Model

Nowadays, workflow technology has become one of the main techniques which supports for the field of office automation. The traditional workflow can not adapt to the distributed operating environment and loosely coupled heterogeneous distribution of information, increasingly nowadays. To solve these problems, with the Agent used in conjunction with the workflow, this thesis use the intelligent features of Agent such as autonomy, interactivity and initiative, which can give consideration to the security issues of workflow system on the basis of improving the traditional mode of operation of the workflow. An improved of Role-Based multi- Agent access control model is presented to provides an effective solution for workflow management system.Based on the analysis of the RBAC model not fully being applicable to the workflow system, the model privilege distribution mechanism is proposed by introducing the Agent technology, task, the instance set of task, task context and other related concepts.The improved model which retaining the advantages of the RBAC model is used in the workflow system, which set the collaborative relationship among users, roles, tasks, permissions and the different types of Agent.It can solve two important problems, the separation of duty and the least privilege,which enhances the security and flexibility of the system.The hierarchical model of operating mechanism are described. The way of information transfer between entities is clearly articulated. The agent types and interactive collaboration, the main module in specific applications and the database relational model are discussed at the same time. Finally, the.example of multi-agent intelligence processing system is given in the platform of WADE. An improved of Role-Based multi- Agent access control model is verified. Through setting the relationship among the roles, tasks and privileges of the system, and strictly limitting of the privileges when users perform tasks, the validity of the model is feasible and verified.