Journal of Software Engineering for Autonomous Systems

Collaborative systems such as cyber-physical systems (CPS) dynamically form runtime networks to achieve goals that cannot be achieved by the individual system alone. While much research has been done on architectures for adaptive systems, these approaches do not consider the fact that goals exist on network level and thus need specific adaptation strategies. In decentralized settings where there is no central mechanism for orchestrating and coordinating the collaboration between systems, failure to adapt individual systems’ behaviors without considering the network’s overall outcome may result in a functional degradation of the entire network, possibly to the point of injuring humans and harming individual systems in the network. In this article, we propose a goal-based adaptation approach that takes the goals of the individual systems as well as the goals of the overall network into account. Our approach allows considering adaptation strategies based on the network goals which are fulfilled by automated behavioral adaptations of the individual system of the network. Results from a proof-of-concept application to an industrial case example show applicability and usefulness of the approach.