Using Failure Semantics to Maintain Safety for Dynamic Composed Systems

Inhalt:
Cyber-physical systems feature the ability to dynamically compose components in accordance to its demands. From this flexibility follows that it is unknown at design-time which system composition will be available at run-time. Since sensors inherently suffer from a lack of precision and accuracy, the lack of knowledge on this failure behavior is in conflict with maintaining safety where a design-time guarantee assures that sufficient fault-tolerance will be provided at run-time. In this paper, we focus on the challenge of dynamically using sensor data while maintaining safety. To manage this, we propose using failure semantics in order to describe the failure characteristics of a sensor as well as the fault tolerance capabilities of an application at design-time. Once the system composition becomes available at run-time, matching the failure semantics provided by the sensor and the failure semantics satisfied by the application enables us to maintain safety even if the system is composed dynamically. Finally, we present our findings in the context of a robotic localization where failure semantics are used to incorporate dynamically observations with respect to given safety requirements.