Title:
Safety Assurance of Cyber-Physical Systems: Mind the Gap

Preliminary Abstract:
CPSs are distributed, heterogeneous and interconnected systems, typically engineered from the seamless integration of computational algorithms, physical processes and communication technologies. While the ability of CPSs to interact with the physical world is vital in realising practical real-word CPS applications, it also reinforces the uncertainty associated with the engineering of CPSs. The use of safety assurance methods at design time aids to tackle some uncertainty facets through systematic argumentation that the CPS will function as intended and instantiation of traces between safety requirements and artifacts spanning the CPS engineering stages, leading to the formulation of an assurance case. Other facets, however, including autonomous decision-making enabled by artificial intelligence algorithms, are primarily manifested while the CPS is in operation with their outcomes not being easy to anticipate. This could lead to a global emergent CPS behaviour putting safe and reliable system operation in jeopardy. We argue that there is a need for developing new or adapting existing safety assurance methods, and devising suitable abstractions that enable software-intensive CPSs to establish a rationale for the decisions being made and to link these decisions with safety requirements, thus providing the means to increase our confidence for trustworthy CPS behaviour.