Date of Award

2019

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

First Advisor

Dr. William W. Edmonson

Abstract

Cyber-physical systems (CPS) are systems that exhibit tight integration between their physical and computational components. They are hybrid systems containing continuous states and discrete states emanating from the physical and computational components, respectively. Systems engineering of these systems is challenging due to the tight integration of their computational, physical and communication technologies. As their level of acceptance increases in mission critical applications such as health care, smart grid, autonomous vehicles and smart cities, the need to ensure their safe operation is paramount importance as well. Therefore, design methodologies followed in their development are required to result in system behavior that is reliable and predictable. In this work, we propose a systems engineering methodology called the Verifiable Design Process (VDP) that if followed properly will result in a correct-by-construction of cyber-physical systems. The VDP achieves correct construction of CPS by integrating formal methods with model based systems engineering (MBSE) approach. We utilize formal methods in the VDP to formally verify CPS mission critical properties against their requirements. Since CPS constitute systems of systems or complex systems, we utilize model based systems engineering approach to manage design complexity as well as provide validation based on modeling and simulation. The VDP consists of hierarchically arranged abstraction layers where which layer contains requirements in natural language form, specification in logic form, systems of models and their associated simulations. The abstraction layers allow divide and conquer approach to the design process. The abstraction layers are related to each other through refinement and abstraction functions. Any systems engineering process begins naturally with a set of requirements in natural language. Natural language are preferred because of its easiness for facilitating communication among multiple stakeholders involved in the project. Unfortunately, natural languages have inherent ambiguities, inconsistencies, incompleteness, and hence result in incorrect requirements. There is a need to formally represent these requirements. In the VDP, these is achieved by 2 mapping set of requirements in natural language to logic based representation. In addition to logic based representation, a system of models are created to enable model based simulation. Model based simulation provides a validation mechanism to stakeholders, thereby ensuring designs are confirmed to represent the true needs of stakeholders at each stage of abstraction. The contributions of this work are the following: 1) Introducing ontological representation of domain requirements; 2) Introducing an ontology driven requirements engineering process in the VDP; 3) Developing formal foundation for CPS design process using category theory; 4) Formalizing the various forms of representation in the VDP (set of requirements, logical specifications, models and their associated simulations) and their relations using category theory.

Download

Share

COinS