Ensuring the dependability of technical facilities through triplication and quadrupling

Abstract

Redundancy, e.g. structural redundancy, is one of the primary methods of improving the dependability, ensures failsafety and fault tolerance of components, devices and systems. According to the International Patent Classification (IPC), the class of systems and methods G06F11/18 is defined as «using passive fault-masking of the redundant circuits, e.g. by quadrupling or by majority decision circuits». Obviously, «fault-masking» masks not only faults, but failures as well. The majority decision circuits (MDC) in the minimal configuration implements a «2-out-of-3» choice. According to the above definition, such redundancy should not require a special decision circuit. However, that is not always the case. In cases when the resulting signal out of a quadruple logic is delivered to, for instance, an executive device, a «3-outof-4» selection circuit is required anyway. Another dependability-improving solution is defined by class G06F 11/20, «using active fault-masking, e.g. by switching out faulty elements or by switching in spare elements». The word «active» is missing here, thus we have active and passive fault tolerance. The paper examines passive fault tolerance that uses triplication and quadrupling and compares the respective probabilities of no-failure.The Weibull distribution is used that most adequately describes dependability in terms of radiation durability under the effects of heavy ions. It shows that in a number of cases quadrupling has a lower redundancy than triplication. A formula is proposed that describes the conditions of preferability of quadrupling at transistor level.