On reliability estimation for combined transient and permanent fault handling

This paper addresses the problem of modeling the reliability of systems, where permanent and transient faults are handled in a combined manner. First, we briefly introduce the motivation for handling permanent and transient faults within a combined architecture. Then, based on the well known TMR scheme, we present the difference in modeling the reliability for permanent and transient faults. Subsequently, we introduce two distinct methods for modeling both fault types together and explain how the calculation of the reliability can be done. Both models vary in their accuracy and complexity. Furthermore, we apply the proposed modeling strategies to a more complex system, where permanent fault compensation is used to keep transient fault detection/masking unaffected. Modeling this system with both methods is discussed for varying system parameters and failure rates. The results show that in particular situations the use of the simple model is sufficient and makes system modeling less time consuming. But in some cases the inaccuracy of the simplified model will cause an excessively optimistic estimation of the reliability.