A Scalable Component-based Modeling Environment Supporting Model Validation

Simulation validation remains a central and growing challenge, especially when models are large-scale and complex. To advance rigorous simulation validation, we have developed a component-based modeling environment called Scalable Entity Structure Modeler (SESM), which is based on the multi-aspect, multiresolution component-based modeling paradigm. Using a multi-view graphical modeling interface, modelers can specify and compose large and complex models with arbitrary hierarchies. Models can be manipulated and stored directly in standardized relational databases—simplifying model specification and simulation experimentation tasks. A key advantage of a persistent model-base is that it enables modelers to measure a model’s structural and some behavioral complexity metrics (e.g. number and types of components, inputs, outputs, and states of models). A realization of this approach called Scalable Entity Structure Modeler with Complexity Metrics (SESM/CM) is developed. It supports syntactic and semantic correctness of domain-neutral properties of models given a set of well-defined relationships among model components and their interactions. The model components in the model-base can be transformed into XML and subsequently simulation model components suitable for simulation environments such as DEVSJAVA. The scalability and transformation features of the approach offer a sound basis for simulation at varying levels of abstraction, including arbitrary yet systematic validation of models in terms of their observable input and output relations and state transitions. Direct support for incremental configuration of simulation experimentations and executions is important for large-scale simulation validation. To this end, the model-base logically lends itself to the management of simulation experiments. This approach supports formal specification of models, design of experiments, and observation of simulation executions. The proposed modeling approach is illustrated using a computer network example. We discuss simulation validation and future research directions aimed at supporting integrative modeling and simulation of enterprise, supply-network systems.