Decision-Centric Foundations for Complex Systems Engineering and Design
Abstract
During the past decade, there has been significant interest in bottom-up evolutionary complex systems, such as the Internet, smart transportation systems and social product development. Compared with hierarchically-designed systems such as automotive and aircraft, the architectures of such complex systems are not under the direct control of designers, but emerge in a bottom-up manner based on decisions made by individual entities. The design strategy for bottom-up evolutionary complex systems is to influence agents' behavior at the micro-level in order to indirectly achieve the requirements and desired performance at the system-level. To this end, the research objective of this dissertation is to establish a framework to model, analyze and estimate the micro-level decision-making behaviors for facilitating complex systems engineering and design. Existing studies have provided insights on modeling micro-level behaviors and understanding their effects on system-level performance. However, there is a lack of theoretical foundations for explaining why agents' behaviors are modeled according to the rationality assumptions and whether or not such assumptions are appropriate. In other words, existing studies are primarily focused on the outcomes of decision-making (micro-level behavior) instead of the reasons for those decisions (decision-making preferences). There is a research gap in understanding the effects of agents' decision-making preferences on system structure and dynamics. To address this research gap, a decision-centric framework is proposed in this dissertation. This framework provides theoretical foundations for explaining agents' rational behavior, establishes a link between decision-making preferences and the micro-level behaviors, and builds the relationship between agents' preferences and system-level performance. Towards establishing the decision-centric framework, three research questions regarding modeling, analyzing and estimating micro-level behavior are addressed. The approaches proposed for answering the questions are validated using two applications examples: the Internet and social product development. In the case of the Internet, an approach based on discrete choice random utility theory integrated with complex network analysis is proposed to obtain micro-level behavioral models. The primary outcome is a generic decision-centric approach for modeling the evolution of complex systems. In the case of social product development, by integrating game theory and behavioral experimentation, a rigorous framework is established for modeling the decision-making behaviors of individuals in crowdsourcing competitions, and understanding its effects on the design outcomes. The overall contribution in this dissertation is a decision-centric framework to support bottom-up engineering and design of complex systems. The approaches established in this dissertation support the attainment of new knowledge on modeling, analysis and estimation of micro-level behaviors in complex systems. The results presented in this dissertation provide insights on directing the design of incentives and mechanisms to influence agents' interactions at the micro-level to achieve desired system-level performance in complex networked systems, and to improve the crowdsourcing process for better design outcomes.
Degree
Ph.D.
Advisors
Panchal, Purdue University.
Subject Area
Behavioral psychology|Mechanical engineering|Systems science
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.