Predictability in Critical Systems

Predictability is crucial in critical applications and systems. Therefore, we examine sources of uncertainty for each of the four phases that span a project lifecycle, from initial problem capture, to system implementation, when conducted according to proof-based system engineering principles. We explore the concept of coverage applied to problems, solutions, assumptions, along with a generic problem that arises with critical applications such as, e.g., air traffic control/management, namely the real-time uniform atomic broadcast problem. We examine two design styles, namely asynchronous and synchronous solutions, and compare the resulting assumptions as well as their coverages. The central issues of overloads and timing failures that arise with synchronous models are investigated in detail.