Automatic Generation of Contractual Requirements from MBSE Artifacts

Abstract

This report describes recent research in support of acquisition programs that leverage requirements as contractual elements. Textual requirements form the backbone of contracting in acquisition programs. Requirements define the problem boundaries within which contractors try to find acceptable solutions (design systems). At the same time, requirements are the criteria by which a customer measures the extent to what their contract has been fulfilled by the contractor. However, current government reports and academic research show latent problems in acquisition programs stemming from poor practices in requirements engineering. In order to cope with such a challenge, academia and industry envision extending the application of MBSE beyond conceptual design, particularly addressing problem formulation. Two main paths to integrate requirements within a complete MBSE environment are currently pursued. In the first path, major modeling languages, such as SysML, incorporate elements called requirement models, which are intended to model the requirements the system is expected to fulfil. However, the only modeling value of this approach is to achieve traceability between requirements and architectural elements. In the second path, researchers propose to use behavioral models of the system of interest as problem definition elements (requirements). However, the proposition remains positional, since such work has not addressed how contracting in acquisition programs is affected, or needs to be adjusted, to incorporate behavioral models as a contractual mechanism instead of textual requirements. Hence, the near-term, practical feasibility of the approach is questionable. In order to cope with these challenges, this research project addressed the main question of whether contractual requirements in textual form can be automatically generated from requirement models in an MBSE environment without loss of information or intent. In particular, this research had the following objectives: (1) Create requirement models that can capture all information and intent of textual requirements; and (2) Translate requirement models into textual requirements without loss of information or intent in terms of contractual needs in acquisition programs. The research employed a combination of theoretical foundations and tool development and implementation. The hypotheses were tested on an Air Force Institute of Technology notional satellite. By fulfilling the research objectives, the results of this research are anticipated to significantly improve the performance of acquisition programs, in particular with regards the generation of contractual requirements. Furthermore, the direct public benefit of this research is anticipated to be higher early efficacy of commercial products and public services. Finally, while we considered an application for the Air Force as a test case, we anticipate that the methodologies and insights provided in this work can be applicable to a broad range of systems that require careful definition of requirements: other defense systems, space systems, aeronautics, automotive systems, manufacturing systems, electronic products, civil infrastructure, public health systems, or transportation systems. The research has already resulted in one published paper for the 2019 Acquisition Research Symposium, one published paper for the 2019 Conference on Systems Engineering Research (CSER), and one published paper in the Systems journal. Several other conference and journal papers resulting from this research are currently under preparation and will be submitted before the end of 2019.