Date of Award

12-14-2017

Degree Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Computer Science

First Advisor

Dr. Xiaolin Hu

Second Advisor

Dr. Robert Harrison

Third Advisor

Dr. Xin Qi

Fourth Advisor

Dr. Ying Zhu

Abstract

Simulation modeling provides insight into how dynamic systems work. Current simulation modeling approaches are primarily knowledge-driven, which involves a process of converting expert knowledge into models and simulating them to understand more about the system. Knowledge-driven models are useful for exploring the dynamics of systems, but are handcrafted which means that they are expensive to develop and reflect the bias and limited knowledge of their creators. To address limitations of knowledge-driven simulation modeling, this dissertation develops a framework towards data-driven simulation modeling that discovers simulation models in an automated way based on data or behavior patterns extracted from systems under study. By using data, simulation models can be discovered automatically and with less bias than through knowledge-driven methods. Additionally, multiple models can be discovered that replicate the desired behavior. Each of these models can be thought of as a hypothesis about how the real system generates the observed behavior. This framework was developed based on the application of mobile agent-based systems. The developed framework is composed of three components: 1) model space specification; 2) search method; and 3) framework measurement metrics. The model space specification provides a formal specification for the general model structure from which various models can be generated. The search method is used to efficiently search the model space for candidate models that exhibit desired behavior. The five framework measurement metrics: flexibility, comprehensibility, controllability, compossability, and robustness, are developed to evaluate the overall framework. Furthermore, to incorporate knowledge into the data-driven simulation modeling framework, a method was developed that uses System Entity Structures (SESs) to specify incomplete knowledge to be used by the model search process. This is significant because knowledge-driven modeling requires a complete understanding of a system before it can be modeled, whereas the framework can find a model with incomplete knowledge. The developed framework has been applied to mobile agent-based systems and the results demonstrate that it is possible to discover a variety of interesting models using the framework.

Share

COinS