Framework for component-based distributed simulation on the grid : specification and coordination.
Date of Issue2009
School of Computer Engineering
Parallel and Distributed Computing Centre
The emergence of Grid computing is a milestone in high performance computing. The Grid computing promotes flexible, secure, coordinated, and massive resource sharing among networks. Component technology boosts code reusability. The service-oriented ar-chitecture is an ideal paradigm for next generation computing. The loose coupling among services in the service-oriented architecture relieves service consumers from detailed knowledge of implementation, implementation language, and execution platform of the services to be consumed. It is innovative and advisable to take advantages of both the Grid computing and component technology and apply these technologies to distributed simulation in a service-oriented architecture. A framework for component-based distrib-uted simulation on the Grid comes into being with the above considerations. The framework facilitates the development and execution of distributed simulation in different views, i.e., a component-based view for development and a service-oriented view for execution. At component-development level, the framework standardizes the simulation components and simulation applications in order for interoperability between components and reusability of the existing components. A simulation component con-forms to a standard component interface, and a simulation application conforms to a pre-defined simulation description schema. At service-execution level, each component is executed as a Web Service in Globus Toolkit version 4. A pre-deployed organizer service is employed to initialize a simulation application, to dispatch the execution-time informa-tion to relevant components, and to collect simulation result. The framework supports shared variable management in distributed simulation by introducing history list and fu-ture list. This thesis details the framework pertaining to the specifications, the design of the ar-chitecture and its components, automatic service generation, runtime implementation of the framework, and different approaches for shared variables management. The profiling on an automatically generated service is carried out. An experiment on the shared vari-ables management is conducted to compare the performances of different approaches in shared variables management. In addition, a case study of wafer fabrication is used to il-lustrate how the above concepts can be applied.
DRNTU::Engineering::Computer science and engineering::Computer systems organization::Computer-communication networks