Model-Based Systems Engineering (MBSE) Methodology for Integrating Autonomy into a System of Systems Using the Unified Architecture Framework
Mohammadreza Torkjazi, Ali K. Raz (George Mason University)
Keywords
Autonomy integration;Level of autonomy;SoS Architecture;MBSE methodology;UAF
Abstract
Many stakeholders of existing Systems of Systems (SoSs) are interested in leveraging the new ca-pabilities provided by autonomous systems empowered by Artificial Intelligence (AI). This requires the integration of these systems into SoSs, resulting in Systems of Autonomous Systems (SoASs). SoAS architecting is different from SoS as the architecting challenges are exacerbated by the Level of Autonomy (LoA). An autonomous system has various LoAs depending on its AI advancement and the capabilities it provides. Each LoA impacts the managerial and technical challenges for SoAS architecting in a different manner. The managerial aspect covers concerns such as policies and agreements, whereas the technical aspect highlights issues such as compatibility between autonomous systems and legacy systems. Failure to address the LoA impact on these factors in the architecting phase results in an ineffective integration. In this paper, we propose a methodology that follows the SoS hierarchical lexicon, builds upon the standard steps of the Object-Oriented Systems Engineering Method (OOSEM), and leverages the Unified Architecture Framework (UAF) for modeling autonomy integration. The proposed methodology adds detailed sub-steps to OOSEM, where we introduce the required UAF views for modeling each aspect of the SoAS architecture. This methodology lays the foundation for the trade study analysis that helps stakeholders decide on suitable LoAs for SoAS. We also present an illustrative example to demonstrate the implementation and effectiveness of the proposed methodology.