Reliability Improvement of Real-Time Embedded System Using Checkpointing

The checkpointing problem in real-time embedded systems is dealt with from a reliability point of view. Transient faults are assumed to be detected in a non-concurrent manner (e.g., periodically). The probability of successful real-time task completion in the presence of transient faults is derived with the consideration of the effects of the transient faults that may occur during checkpointing or recovery operations. Based on this, an optimal equidistant checkpointing strategy that maximizes the probability of task completion is proposed.