Research On Multi-Agent Oriented Policy Representation Framework And Policy Conflict

Policy is firstly used in the areas related to computers and networks in the form of rules, strategies and protocols. These areas include database security, distributed network and mobile network technologies. However, from the standpoint of policy management, only the functions at lower layer in the policy management are realized. That's to say, only a kind of restricted management to the system execution is realized which can be considered as passive management. In the multi-agent environment, the passive management can't meet the requirements of applications to enable the intelligent agent to show the capability of self-coordination and self-management, therefore, a new policy representation framework is built in this thesis. The framework not only can restrict the behaviors of intelligent agent, but also can guide, drive the intelligent agent to make reasonable decisions and select the proper behaviors to realize proactive management of the policy.Meanwhile, it's unavoidable that some semantic conflicts will happen in different policies, so detecting and harmonizing conflicts is becoming a must work after setting up the policy representation framework. However, the current researches on policy conflicts are not enough from standpoints in the breadth or depth. In the breadth, the types of policy conflicts are relatively unitary, and a comprehensive consideration is not made to stages where the policy conflicts may happen. In the depth, although some detecting and harmonizing strategies for policy conflicts have been proposed, most of them are designed for specific systems. In this thesis, our research on policy conflict is based on a new kind of policy representation framework, multi-layer, multi-type and multi-phase policy conflict types and the corresponding detecting and harmonizing algorithms for policy conflict are proposed.Finally, in order to simplify policy design work and in the combination of usability, functionality and effectiveness, a special policy management tool is developed to encapsulate the complexity of policy design work and ease the user operation.In one word, the main work and contributions are shown below:1. A kind of layered policy representation framework is proposed. The Framework embodies the driving ability of the policy to the behaviors of the intelligent agent and realizes the proactive management of the policy. The policy representation style which combines the object-oriented and ontology theory is used to assimilate the advantages of ontology and object-oriented in policy representation to make up for the deficiencies of the two.2. Many kinds of policy conflict types and related algorithms based on the new Policy representation framework are proposed which covers the whole phases where policy conflicts may happen and provides a new topic for research on policy conflict.