A Fault Tolerant Scheduling Algorithm for Stochastic Fault Model in Real-Time Operating System

For real-time system, fault tolerant scheduling algorithm is an important method to guarantee the timing constraints of tasks when fault occurs. In the time before, researchers usually describe faults by assuming a constant number of faults or the minimum time of fault inter-arrivals. In practice, occurrence of faults of systems are stochastic. In this paper, we model the occurrence of faults as a stochastic process with a Poisson distribution having a mean inter-arrival rate of ¿. And a fault tolerant task scheduling algorithm which is a greedy algorithm with a lower complexity of computation is given according to the fault model. Through simulation we can conclude the algorithm is effective for fault tolerant. The loss ratio of tasks on this algorithm is much lower than that in no fault scheduling algorithm.