Research On Automatic Subscription Decomposition Model And Matching Algorithm In Publish/Subscribe Systems

The emergence of new network computing model,such as Mobile computing,grid computing,cloud computing and pervasive computing,has led to great changes in the scenarios of distributed application systems'.In the traditional distributed systems, each node exchanges information and communicates with other nodes using request/response method.This synchronous communication model is difficult to adapt to a new generation of distributed computing environment that is characterized by the large scale,loosely decoupled and scalability.The Publish/Subscribe systems have abilities to make information producers and information consumers completely decoupled in time and space that meets the needs of the large-scale information distributed systems,so it has gradually become the infrastructure platform for the large-scale distributed systems.However,the existing Publish/Subscribe systems do not have abilities to support semantics heterogeneous.In some applications,an atomic subscription cannot get the events that the subscribers are most interested in or even cannot get the match result. To address these issues,supported by National 863 program and National Natural Science Foundation,we research on the data model of Publish/Subscribe systems. trying to implement automated subscription decomposition supporting semantic equivalence and matching algorithm and then implement a prototype system JTangPS-D verified the proposed model.Fist,we propose a data model that can implement automated subscription decomposition supporting semantic equivalence.Ontology is introduced into Publish/Subscribe system to build conceptual model,expressing all of the concepts and relationships between them.RDF can be used to exchange information between applications and guarantee not to change the semantic.In order to address the issue of semantic heterogeneous,we use RDF graphs to represent events and RDF graph patterns to represent subscriptions,and design a subscription language PSSL.Then.we define component semantics.On this basis,an automated subscription decomposition algorithm supporting semantic equivalence is designed according to decomposition model.The decomposition algorithm decomposed the automated subscription into composite subscriptions,thus increasing the system's express ability and usability.An efficient semantic-based matching algorithm is proposed based on the data model.We save all of the system's semantics information into the RDF Node Index Structure based on the subject's types of the subscriptions' triples.The same subscription conditions between different subscriptions are saved only once.The matching process is to traverse the event graph node;find the subscriptions' nodes which match with the event node in the semantic,the syntax and graph structure; update the matching status of each subscription,and then proceed matching on the termination node of event arc.At last,the algorithm returns the set of matching subscriptions.Finally,we design and implement a prototype system JTangPS-D.