B-space: dynamic management and assurance of open systems of systems

Abstract

Connected cars, freely configurable operating rooms, or autonomous harvesting fleets: dynamically emerging open systems of systems will shape a new generation of systems opening up a vast potential for new kinds of applications. In light of the hard-to-predict structure and behavior of such systems, assuring their safety will require some disruptive changes of established safety paradigms. Combining current research results from different disciplines with industrial experience, this paper dares to think out of the box and look beyond the limits of traditional safety assurance. It structures upcoming challenges posed by the emergence of open systems of systems, tries to shift existing paradigms to meet those new challenges, and proposes an abstract conceptual framework building on comprehensive interlinked multi-concern runtime models for dynamically assuring the safety as well as other properties of open systems of systems. As there currently is no comprehensive realization of the framework, we discuss what kind of approaches could fit into which parts of the framework and exemplify this for the case of conditional safety certificates.

Defining and Researching “Dynamic Systems of Systems”

“…If essential safety functions are realized by a dynamic composition of (autonomous) systems, then it is challenging to assure that the essential safety functions are always provided in sufficient quality. Research addressing this issue is referred to as runtime certification [37] or dynamic risk management [38].…”

An Assurance Case Description Method to Reduce Misunderstanding of Managed Objects When Designing Open System-of-Systems

“…Additionally, to define a safety assurance approach for open SoS, the previous study suggests a classification that primarily reflects those classification criteria that are considered for safety assurance. However, the previous study [30] states that as of today it is not clear how the vision can and will be achieved. Therefore, the above previous studies do not focus on describing ACs to reduce misunderstanding of managed objects when designing open SoS.…”

“…This subsumes a brief discussion of the relationship between ConSerts and traditional safety engineering activities, the consideration of some prerequisites for the suitability of the approach, and detailed information on the specification, transformation, and dynamic evaluation of ConSerts. The previous study [30] introduces B-Spaces, which is called virtual space, as a conceptual framework for comprehensive multi-concern and multi-level models at runtime. To structure the different kinds of models@run.time [31] that can be used in a B-Space, the previous study [30] utilizes three dimensions that reflect the purpose of the model.…”

“…The previous study [30] introduces B-Spaces, which is called virtual space, as a conceptual framework for comprehensive multi-concern and multi-level models at runtime. To structure the different kinds of models@run.time [31] that can be used in a B-Space, the previous study [30] utilizes three dimensions that reflect the purpose of the model. Additionally, to define a safety assurance approach for open SoS, the previous study suggests a classification that primarily reflects those classification criteria that are considered for safety assurance.…”