BOTTOMS-UP: A SYSML APPROACH TO REVERSE ENGINEERING CYBER-PHYSICAL SYSTEMS

Abstract

Cyber-Physical Systems are often examined through reverse engineering, particularly when original design documents are unavailable, outdated, or proprietary. However, reverse engineering is rarely integrated into Model-Based Systems Engineering workflows, creating a disconnect between low-level technical findings and the broader system design intent. This thesis presents a practical methodology for reconstructing partial System Modeling Language models directly from reverse-engineered artifacts. The approach treats the modeling language not only as a verification framework but as a means of expressing inferred design structures, behaviors, and requirements, and then iteratively verifying them with reverse-engineering. Through case studies on embedded systems, this research demonstrates how evidence gathered from reverse engineering, such as interface analysis, communication traces, and code examination, can be systematically represented within behavioral, structural, and requirement diagrams. By incorporating hypothesized requirements and real observed mechanisms, the resulting models provide a coherent view of how system functions emerge and interact. Traceability relationships such as satisfy, verify, trace, threaten, and mitigate are then used to link reverse-engineered artifacts back to inferred design intent. These models support reasoning about system behavior, architecture, and security implications in a structured and transparent manner. This thesis establishes a repeatable process for reconstructing those models and analyzing complex systems, enabling deeper technical understanding even in the absence of formal specifications.