From users' perspective, there are two system usefulness states: service accomplishment and service interruption. By service accomplishment, we mean that service is delivered as the specified whereas by service interruption, delivered service differs from the specified. In case of system failure, the system usefulness state switches from the service accomplishment to the service interruption. Hence, efforts must be made to limit the system failure. However, if system failure could not be wholly prevented, some means to bring the system from interruption state back to accomplishment state are necessary.
Radiation-induced transient faults, a .k. a. single-event upsets (SEUs), are one of run time causes that could end up incorrectness of the microprocessor executions. Hence, we are developing mechanisms, mostly at architecture and/or microarchirecture levels, to protect microprocessors from SEUs. However, to that end, the first step must be a systematical examination of the SEU issue to gain the fundamentals.