The application of Agile to large-scale, safety-critical, cyber-physical systems

Abstract

The increasing complexity of large-scale, safety-critical cyber-physical (LS/SC/CP) systems, characterized by interconnected physical and computational components that must meet stringent safety and regulatory requirements, presents significant challenges to traditional development approaches. Traditional development approaches, such as the waterfall methodology, often struggle to meet adaptability, speed, and continuous assurance demands. This dissertation explores the feasibility of applying and adapting Agile methodologies to LS/SC/CP systems, focusing on challenges like regulatory compliance and rigorous verification, while intending to prove benefits such as improved risk management and faster development cycles. Through case studies and simulations, this research provides empirical validation of Agile's effectiveness in this domain, contributing a framework for adapting Agile practices to meet the unique demands of LS/SC/CP systems. Employing a mixed-methods approach, the research comprises five key components. First, a systematic literature review (SLR) was conducted to assess the current state of Agile adoption in LS/SC/CP environments. Second, a comparative analysis of the top 10 Agile scaling frameworks was performed to evaluate their suitability for LS/SC/CP system development. Third, a survey of 56 respondents provided both quantitative and qualitative insights into industry trends, adoption patterns, and Agile's impact on LS/SC/CPs. Fourth, 25 one-on-one interviews with industry practitioners further explored the challenges, benefits, and enablers of Agile adoption in these environments. Finally, lifecycle modeling (LML) using Innoslate was utilized to develop a fictional case study, modeling the development of a mid-size low Earth orbit (LEO) satellite using both NASA's Waterfall approach (Phase A-D) and an Agile approach with a series of Minimum Viable Products (MVPs). Findings reveal that Agile methodologies, when adapted for LS/SC/CP systems, enable accelerated development cycles, reducing development time by a factor of 2.5 compared to Waterfall while maintaining safety and regulatory compliance. A key contribution of this study is the introduction of a Continuous Assurance Plugin, which integrates continuous validation within Agile's iterative processes, effectively addressing compliance and safety requirements traditionally managed through phase-gated reviews in Waterfall. Additionally, this research provides: 1. Empirical validation of Agile Scaling Frameworks and their suitability for delivering LS/SC/CP systems. 2. Quantitative and qualitative analysis of Agile's current state and impact in LS/SC/CP environments. 3. Evaluation of key enabling technologies such as Model-Based Systems Engineering (MBSE), Digital Twins, and Continuous Integration/Continuous Deployment (CI/CD) that facilitate Agile adoption for LS/SC/CP systems. This dissertation advances the understanding of Agile's role in LS/SC/CP system development, providing actionable insights and practical adaptations for organizations seeking to implement Agile in complex, safety-critical domains.