System observation of blocking, non-blocking, and cascading events for runtime monitoring of real-time systems

The complexity of multitasked applications in real-time embedded systems presents key challenges in reliability of task execution. Interactions between periodic and aperiodic tasks can incur unpredictable deviations from ideal execution times. Runtime observations can provide visibility for analyzing real-time execution behavior of vulnerable tasks and detect when such deviations may lead to system failure - potentially allowing correction or failsafe mechanism to be utilized. However, such runtime observation must be non-intrusive as even small perturbations to the system execution can significantly impact the system. This paper presents a system observation framework for monitoring and analyzing rapidly occurring software events at runtime. The system observation framework provides runtime support for defining and controlling software events with using blocking, non-blocking, and cascading event probes.