Distributed reconfiguration and recovery in the advanced architecture on-board processor

The reconfiguration and recovery approach employed in the advanced architecture on-board processor (AAOP), a fault-tolerant multiprocessor for space applications, is presented. The AAOP is designed to accommodate large numbers of processing elements organized in a distributed fault-tolerant system. The operation of distributed reconfiguration is discussed, and the recovery time is analyzed. Performance of the reconfiguration algorithm under inconsistent observer conditions and against multiple faults is considered. the chordal skiplink ring topology employed in the AAOP is analyzed with respect to its node-pair distance distribution as a function of the number of faults injected. Extensions of this topology are also considered. These would reduce the network diameter and increase fault robustness but also increase the network management overhead.<>