A proposal for fault-tolerance in a multiprocessor system

In this paper, a fault-handling approach for distributed architectures is presented by referring to a specific multimicroprocessor system, i. e., the MuTEAM prototype. The approach is intended to take into account the following hardware and software specifications: i) the system is not provided with mass storage devices; ii) its activity consists of running a set of concurrent processes, cooperating through message exchanges; and iii) no erroneous output is generated toward the external environment even when a fault is present.