BEHAVIORAL MODEL SPECIFICATION TOWARDS SIMULATION VALIDATION USING RELATIONAL DATABASES

Many contemporary systems that are inherently large-scale and complex can be specified using system-theoretic and object-oriented modeling concepts and principles. To examine these systems via simulation and in particular in terms of model validation, it is important to use model repositories. The structure and behavior of dynamical systems can be represented as atomic models having inputs, outputs, states, and functions. Scalable System Entity Structure Modeler with Complexity Measures (SESM/CM) offers a basis for developing modular atomic and composite simulation models as well as non-simulatable models. It allows simulation models to be stored, retrieved, and managed in relational databases. The environment, however, does not provide capabilities for characterizing and storing behavioral aspects of models or their transformation for simulation execution. This thesis describes a design and implementation for capturing some behavioral aspects of atomic models and the transformation of the models captured in SESM/CM into their compatible simulatable formats in DEVSJAVA a simulation environment capable of executing discrete-event models. The combined modeling and simulation capability offers a process where users can develop models in the extended SESM/CM modeling environment and validate their behavior using the DEVSJAVA simulation environment. The proposed extensions to the SESM/CM are demonstrated using a simulated anti-virus computer network model.