Online error detection
Detection of violations of system specifications at run-time enforces dependability, since it may be used as a trigger for possible recovery actions.
In this work, we focused on run-time monitoring of system properties that include both causality and timing constraints, in distributed and time-constrained systems. Based on a description of a property that includes events and temporal constraints, expressed as a timed-arc Petri net, we developed an automated transformation that translates it into an executable and distributed monitoring engine.
We introduced a modification of the semantics of Petri nets [BFR12, BFR13] to be able to execute it online on partial executions and distributed observation environments. Based on this formal framework we developed Minotor [BFR13], a model-driven distributed monitoring system, that is able to express advanced specifications that describe several system operation modes, i.e., nominal mode, warning mode and degraded mode, as shown in Figure XX.
The synthesis of the monitoring engine from the specification results in a lightweight and fully distributed program that is integrated with the system to be monitored. Hence, our approach is not limited to control-and-command systems, and we are investigating its integration into a wireless sensor network application.
A railway supervision system and its associated Petri net for monitoring
(green: nominal mode, red: degraded mode, blue: warning mode)
[BFR13] O. Baldellon, J-C. Fabre, M. Roy: Minotor: Monitoring Timing and Behavioral Properties for Dependable Distributed Systems. The 19th IEEE Pacific Rim International Symposium on Dependable Computing (PRDC 2013), Vancouver, British Columbia, Canada, 2-4 December 2013
[BFR12] O. Baldellon, J-C. Fabre, M. Roy: Distributed Monitoring of Temporal System Properties Using Petri Nets. IEEE SRDS 2012, pp.398-399