An Efficient Hardware Fault-Tolerant Technique

In this work, we introduce a new concept called the dynamic group maximum matching. Under this concept, the system is partitioned into disjoint groups with different sizes dynamically. Then, processors of each group are assigned to run the same task and output of the task is obtained by comparison among the outputs of the processors within the group. The group size is proportion to the task reliability requirements. We introduce a Dynamic Group Maximum Matching (DGMM) algorithm for finding groups dynamically. We also propose an efficient hardware fault-tolerant technique for reliable execution of tasks and concurrent diagnosis of faults while processors and communication channels are subject to failures. Furthermore, we study the effect of the proposed technique on the system performance.