Research On Key Technologies Of Service-Oriented Simulation Supporting Environment

Simulation Supporting Environment(SSE) is an environment where services with no direct relation to the simulation system functionality but with the performance of several other operations necessary for the rest of services to operate correctly are provided. As a fundamental environment for simulations, SSE makes sure needs can be satisfied with suitable resources, thus its major functions are resource management and useage support for simulations. Nowadays, due to the development of web technologies,simulation resources in different organizations, domains and locations can be shared and reused, which greatly cuts the cost and time for simulation development. However, it also brings complexity to the characteristics of resources, such as belongingness, type,authority and availability, thus makes it more difficult for SSE to manage resources.Meanwhile, the requirements for the SSE in efficiency and orchestration are even higher,presenting great challenge to the traditional simulation technologies.With inherent support for interoperability and dynamic composability, Service Oriented Architecture(SOA) can well support fast deployment and orchestration, thus it becomes the trend for the next generation of SSE. We studied the key techniques in realization of SSE based on SOA in this paper. We investigated the resource sharing architecture, resource discovery method and the mechanism of simulation automation respectively, where the resource sharing architecture is the backbone for SSE; resource discovery method and the mechanism of simulation automation are key technologies for SSE, guarenting resources assigned properly at right time.Main work and innovations of this dissertation include:1) We propose a novel Cloud Computing architecture called sharing among clouds architecture based on Virtual Organization(VO) as the simulation supporting architecture. It is based on hybrid cloud computing architecture and adopts the VO concept from Grid computing to manage resources according to their types and belongingness. This architecture inherits the resource deployment and outsourcing patterns from Cloud Computing, and meanwhile advances in safety, scalebility, agility and robustness. Thus it solves the problem of sharing resources among different cloud vendors;2) We propose the concept of Discovery Oriented Simulation Service Description(DOSSD) and DOSSD Ontology(DOSSDO). According to the specific needs for simulation service discovery, we formulate the DOSSD in a hybrid description manner,which combines the explicit information and implicite information of a resource. We also define the DOSSDO which is the basis for service discovery;3) We propose a DOSSDO based Hybrid Matching(DOSSDO-HM) algorithm,which combines qualitative analysis and quantative analysis to improve the precision ofthe result. DOSSDO-HM algorithm integrates methods of logic reasoning on implicite information and satisfactory evaluation on explicite information to judge whether the service can satisfy the simulation needs. This method improves precision without affecting the recall.4) We propose an optimized method for large scale service discovery. By analyzing services based on ontologies at the registry period, we map the description information of a service to relation database according to its description module. Since online resource discovery are carried out by a simple string-match method, semantic discovery effect can be achieved with a much better performance in efficiency.5) We propose a scenario driven simulation experiment method. Since simulation scenario is the bridge between design and running of a simulation experiment, we build a 5”W”(Who When Where What Why) data exchange model based on it and design a three-tier communication module for semantic interoperation; Based on scenario we also build a simulation lifecycle model to control simulation experiment, which is expecially helpful in a net-centric environment.6) We have designed of the prototype of SSE under the background of simulation for analysis and estimation. We realized parts of the system by the Globus Toolkit 4.The designs of resource registry and mangment module, resource discovery module,simulation management module and resource scheduling module are described in detail.A data structure based simulation case is also designed for the special needs of data management of analysis simulation.