Deadline fault tolerance in a networked real-time system
Peng Paul Wen, Yan Li · University of Southern Queensland ePrints (University of Southern Queensland) · 2006
This paper applies a well-developed algorithm in real-time computing to a networked real-time system, and model the system as a periodic real-time computing one.In this model, each process is named as a task and implemented in two versions: the primary and the alternate.The primaries might fail but the alternates are guaranteed.A scheduling algorithm manages to execute all primaries if possible.Otherwise it guarantees the executing of each task either the primary or the alternate before their deadlines.The algorithm is verified in simulation.The result shows that in cases with high failure probability, the low priority tasks have a lower completion rates.In cases with low failure probability, both high priority and low priority tasks can be completed very well.