Mobile Agent System Design And Implementation

Mobile Agent paradigm is a new programming paradigm for distributed applications. Compared with Client/Server paradigm, what are transferred on the network in this new paradigm are mobile agents' codes and states rather than simple request/reply messages. A mobile agent is a programmable entity having its own thread that can be suspended on the source and resumed on the destination. The key characters of a mobile agent are its mobility and interactivity. Mobility is its essential attribute, and effective interacting is the purpose of moving. Mobile agent paradigm accords better with our thinking and conducting methods in many distributed applications, such as electrical commerce, network management, information retrieval and workflow. Acting on behalf of its owner, an agent who can move is always more popular than a stationary one.The structure design, the object model design, the algorithm design and the" prototype implementation of TechAgent, a pure Java-based mobile agents system, are presented in detail. TechAgent provides an acceptable runtime environment for mobile agents and implements basic concepts and structure of mobile agents system. Using Java 1.1, a breakpoint restart algorithm for agent's thread is given, which not only simulates the thread migration functionally but also gets rid of complexity and inefficiency of the traditional method due to the capture of the runtime stack. The algorithm for plan of agent's tasks separates task-performing codes with agent's kernel codes, enhancing the reusability of agent's kernel codes. Various plans of tasks can be constructed by means of chaining connection primitives and operation primitives the algorithm provides.The prototype implementation of TechAgent is a usable mobile agents system, providing a good experiment environment for research of mobile agents. The prototype can be used as the basis of further research and commercialization of mobile agents system via improving system structure, extending system functions and optimizing system performance.