Integration of adaptable fault management techniques into a dependable middleware architecture

Distributed automation systems such as the one used for the control of electric power infrastructure needs to fulfil requirements such as availability, reliability, safety, fault tolerance, real-timeliness, etc. Those systems have to be able to tolerate changes in the environment and faults. They rely on monitoring mechanisms to identify changes in the resources or the environment to make the appropriate trade-offs (e.g. from cost point of view) for a fast reaction or reconfiguration. This paper presents the integration of a resource monitoring mechanism into a dependable middleware architecture. A mathematical model generates a composite indicator for system performance based on the different monitored parameters. The composite indicator allows dynamic switching of the recovery strategies based on the current state of the distributed system environment. The monitoring mechanism can also be used for proactive fault management techniques.