How can a smart cyber-physical system validate its run-time adaptation actions before and after executing them?

Time has come for engineered systems to behave smartly. Towards this end, they need (i) to collect data directly from real life processes, (ii) to build situation awareness, (iii) to reason about their operational states, (iv) to determine the best servicing objectives, (v) to plan their run-time adaptation, and (vi) to provide dependable operations/services even under dynamically changing circumstances. Designers have to figure out how to design smart cyber-physical systems (S-CPSs) for adaptation at run-time. Designing such S-CPSs is a challenging task. S-CPSs should not prepare only device run-time adaptation plans, but also confirm their feasibility and efficiency. S-CPSs should predict the physical and computational resources. The theory and development methodology of smart systems is still under development. This paper focuses on run-time adaptation, provides an overview on the research efforts, and analyses the results published so far. The literature informs about the fact that there is a need for a meta-model of systems’ self-adaptation, which might however be completely different depending on the kind of systems and the applications. Therefore, the paper proposes that managing self-adaptation decomposes to four logical stages: (i) planning self-adaptation, (ii) verification before self-adaptation, (iii) operationalization of self-adaptation, and (iv) validation of self-adaptation. (Less)