Design And Application Of Action Strategy Language And Decision Engine Oriented To Performance Optimization

The policy-based system management is one of the most efficient approaches to manage large and complex networked systems with a high performance. The key issue of the policy-based management is to design a policy language and its related policy decision engine. IETF has started the policy modeling project since1998. Later on, several policy languages and decision engines are proposed to adopt the policy-based management into scenarios like network management and access control. However, current research for performance optimization of the policy decision engine is insufficient. Problems like inefficient retrieval of policies written in existing policy languages and low response speed of policy decision engine have become the bottleneck of the overall performance guarantee. Thus those problems have stood in the way of the usability improvement of policy languages and decision engines.To tackle problems in policy languages and decision engines, this paper proposes a policy language and a decision engine which are performance optimization-oriented. The policy language is a structured XML-based action policy language, whose expressiveness can cover the device management and the access control management. It is written in the XML standard and adopts the most commonly used ECA paradigm to express its rules. Last but not least, it introduces an index tag to add on a performance optimization-oriented feature. The decision engine designed in this paper can automatically retrieve, analyze and leverage action policies written in the above mentioned policy language. It can also combine the real time environment conditions into the decision result reasoning. Policy conflict detection and resolution methods are realized in the decision engine, and policy indexing, policy caching and tool improvement methods are also introduced to optimize the performance. At the end of this paper is the implementation of the policy language and decision engine in an agriculture knowledge base to support the decision making in the scenario of agricultural Internet of Things. Besides, a series of comparison experiments demonstrate the performance enhancement methods designed in this paper are feasible and effective, and the optimization result become even better when the scale of the knowledge base expands.