Service Oriented Virtualization Technology For Medical Image Grid

Medical Image Grid (MedImGrid) is one of the important applications of grid technology. Currently, the heterogeneities, distribution and dynamicity of web services prohibit WSRF (Web Service Resource Framework) of grid to exert its excellences. In this paper, we proposes the SOV(Service Oriented Virtualization) technologies for MedlmGrid, which uses web service as the joint-point of grid, semantic and virtualization technology, to meet the demands of medical information technology.The SOV research of MedlmGrid aims to shield the application complexity and grid infrastructure complexity, which includes SOV framework, service interoperation model, service selection model and service based program model.The service oriented virtualization layered-framework of medical image grid creates one virtual service layer to realize the combination of grid platform and virtualization technology. It adopts one VCA (Virtual Client Access) based service encapsulation schema and creates layered-metadata and their relations for all layers to improve the management efficiency of resources. To solve the needs to create composite service on demand, the programming model of application layer is also designed for MedlmGrid.The semantic based heterogeneous service interoperation model is designed for MedImGrid: it creates the semantic environment based on hybrid ontologies schema, which is used to shield the heterogeneities of service objects and support the interoperations of web services. The services operations are band with semantic tags to denote the ontology corresponding with current web services. That will be used to direct the semantic mapping between web services messages.The semantic based approximately optimal service selection model is created for MedImGrid: it adopts one semantic based service-view creation schema that includes ontology, semantic context and semantic crossing based service clustering means. The ontology based means realizes the service clustering based on the concept relations of ontology. The semantic context based means combines the user and web service context to predict the possible QoS of acquirable web services and cluster the web services based their QoS. And the semantic crossing based clustering is the combination of above two means to guarantee the service selection is approximately optimal.The semantic workflow based service combination model is proposed for MedImGrid: it defines business, semantic and service workflow; and uses the semantic workflow as the middle layer to simplify the transformation of user demands to service workflow and support the on-demand combination of services. To meet the demand of service-independent service workflow, we design one KNBC (Key Nodes Based Combination) algorithm, which limits the possible errors of web-service execution within little segments of service composition graph to support fault tolerant.The feasibility of MedImGrid SOV is verified through three applications including medical information integration, 3D reconstruction based on CT (Computing Tomography) and medical image retrieval. SOV can help MedImGrid to shield the semantic gulf, improve resource efficiency and help to break through the limits of tightly-coupled resource mode.