Cyber-physical systems (CPS) are often customized to meet customer needs and, hence, exhibit a large number of hard-/software configuration variants. Consequently, the processes deployed on a CPS need to be configured to the respective CPS variant. This includes both configuration at design time (i.e., before deploying the implemented processes on the CPS) and runtime configuration taking the current context of the CPS into account. Such runtime process configuration is by far not trivial, e.g., alternative process fragments may have to be selected at certain points during process execution of which one fragment is then dynamically applied to the process at hand. Contemporary approaches focus on the design time configuration of processes, while neglecting runtime configuration to cope with process variability. In this paper, a generic approach enabling context-aware process configuration at runtime is presented. With the Process Query Language process fragments can be flexibly selected from a process repository, and then be dynamically injected into running process instances depending on the respective contextual situations. The latter can be automatically derived from context factors, e.g., sensor data or configuration parameters of the given CPS. Altogether, the presented approach allows for a flexible configuration and late composition of process instances at runtime, as required in many application domains and scenarios.
Presentation at the 24th Int'l IEEE Enterprise Distributed Object Computing Conference (EDOC’20),
Klaus Kammerer, Eindhoven, Netherlands, 07 October 2020, 2:00 PM (Virtual Event)