The Master-Slave Mobile Agent Model And Its Applications In Network Management

As an important type of Software Agents, Mobile Agent has some key features like mobility, autonomy, pro-activity, collaboration, and intelligence, which will give some new means to the distributed computing. Currently there exist two main problems in Mobile Agent researches: 1, The security of Mobile Agent systems can hardly be guaranteed and cannot meet the needs of large-scale commercial applications. 2, Mobile Agent systems lack some unified formalized description means.Aiming at the previous two problems, the paper made a deep study at the Mobile Agent technology, including especially the Mobile Agent structure and the formalized description means of Mobile Agent systems. The paper then innovatively raised a model called the Master-Slave Mobile Agent Model. In this model, the paper defined two different types of Mobile Agent, which are Master Agent and Slave Agent. The former one can only move physically, while the latter one can move virtually and receive services provided by the former one. The model strictly distinguishes the situations when using mobility or using direct communication, emphasizes the importance of mobility in Mobile-Agent-based systems. The model can save the network resources and simplify the design of Mobile Agent Systems. The paper used the Mobile Ambient Calculus to formally describe the model and reason the behaviors of the model, and made the design of Mobile Agent Systems become standardized and convenient.The Master-Slave Mobile Agent Model is clear and easy to understand, it can help the construction of large-scale distributed applications. Then the paper used the model in network management, gave a network management scheme based on the Master-Si ave Mobile Agent Model, designed and implemented a prototype system based on the scheme. To compare the network management scheme based on the Master-Slave Mobile Agent Model with the traditional schemes, the paper did some theoretic analyses and experiments in the real network environment. The results of these analyses and experiments showed that the network management scheme based on the Master-Slave Mobile Agent Model overcomes the traditional network managementschemes regarding the network resource consumption and the echo time.