Abstract. In software fault-tolerant module, the key issue that affects the performance of fault-tolerant scheduling algorithm is how to predict precisely whether a primary is executable. In order to improve the prediction precision, a new algorithm named DPA, Deep-Prediction based Algorithm, is put forward.DPA uses prediction-table to predict whether a primary can be completed before it is scheduled. Simulation results show that DPA provides more execution time for primaries and wastes less processor time than the wellknown similar algorithms.

No abstract