Peng Junjie1,Hong Bingrong1,Yuan Chengjun2,Wei Zhenhua1,Qiao Yongqiang2
(1.School of Computer Science and Technology, Harbin Institute of Technology, Harbin 150001,China;2.No.771 Institute of Astronautic Science and Technology Group,Xi'an 710054,China)
Abstract:Instruction-retry is one of the most simple and effective methods to recover the onboard system from error state.Unfortunately, how to select reasonable retry time and backup resource during the process of instruction retry has not been resolved satisfactorily.A new retry recovery strategy for single fault,which is based on minimizing the fatal failure probability when the system recovers from fault state, was proposed according to the characteristics of onboard systems. On the premise that all faults occurred in onboard system are single fault, the relationships among instructionretry time, resource backup and the fatal failure probability of the system were given. By minimizing the fatal failure probability, the optimal instructionretry time can be obtained under the condition that the resource backup is set. Meanwhile,if there is a maximum fatal failure probability,the minimum resource backup can be attained when the instructionretry time is optimal. After analyzing the strategy, it shows that the fatal failure probability increases with extending of running time of the system, and the reliability of the onboard system increases with increasing of fault detection probability.
Keywords:onboard system; reliability; instruction retry; fault recovery