Systems engineering is the discipline responsible for assuring the overall success of modern, often software-intensive, engineering projects. Success ultimately hinges on the creation of value for all success-critical stakeholders. Creating value requires understanding, specifying, managing, validating, and evolving complex trade-offs across properties of systems, and linking such properties to stakeholder value. Such properties include, for instance, affordabilty, reliability, evolvability, safety, and confidentiality. Today, however, major projects continue to go awry in large part due to a failure to adequately specify and manage such properties and trade-offs. A core problem is that many properties are defined and understood so imprecisely that they cause miscommunication and confusion. A long history of attempts to use natural and quasi-mathematical language to define such properties has not solved this problem. This project aims to develop and evaluate the highly expressive mathematical, logical, and computational language of constructive logic to produce and validate precise models of system properties, trade-off spaces, and linkages between such properties and stakeholder values. From such formal language specifications, computer implementations of the models will be automatically generated and provided as web services for application and testing by systems engineers.
Formal languages have the potential to clarify and reveal conceptual errors in imprecise, ambiguous, incomplete, and inconsistent natural language specifications. Type theoretical notations offer high expressiveness based on higher-order logic, polymorphic types, and the seamless integration of computational and propositional content, as well as the ability to extract certified software from specifications. This project will employ constructive logic to develop, test, and refine definitions and models of system properties, trade-offs, and the mapping of trade-off spaces to stakeholder preferences. This approach is novel in systems engineering, and important gaps in knowledge remain. To fill them, the following research problems will be addressed (1) determine how best to represent uncertainty in type theoretic definitions of probabilistic system properties; (2) determine appropriate mechanisms for structuring general property theories that can reliably be specialized to specific projects; (3) develop and evaluate languages and models of system properties, including evolvability properties, in particular; (4) develop and evaluate formal frameworks linking system design, property, and value spaces; (5) enable user communities to engage with such theoretical constructs through web services. The project will develop and test the hypothesis that theories based in constructive logic can help systems engineers to reason better about system designs, properties, trade-offs, and stakeholder value.
