Title :
Measuring Reliability as a Mean Failure Cost
Author :
Mili, Ali ; Sheldon, Frederick
Author_Institution :
NJIT, Newark
Abstract :
The Mean Time to Failure (MTTF) is a commonly accepted measure for system reliability. Some variations of it (MTTD: mean time to detection of a vulnerability, and MTTE: mean time to exploitation of vulnerability) have also been adopted to measure other dependability attributes (Stevens et al., 2004) (most notably security). In this fast abstract we submit a tentative challenge to this measure, propose an alternative, and briefly illustrate it with a simple example. When we say that a system S has an MTTF M, we mean that the mean time to the failure of the system with respect to some implicit specification R is M. We propose a measure of reliability that takes into account the variance in failure cost from one supspecification to another, the variance in failure probability from one subspecification to another, and the variance in failure cost from one stakeholder to another. We refer to this measure as the Mean Failure Cost, and abbreviate it by MFC.
Keywords :
formal specification; program diagnostics; software reliability; Mean Failure Cost; Mean Time to Failure; dependability attributes; failure probability; mean failure cost; system failure; system reliability; system specification; vulnerability detection; Aggregates; Carbon capture and storage; Costs; Out of order; Reliability engineering; Security; Sorting; Systems engineering and theory; Time measurement;
Conference_Titel :
High Assurance Systems Engineering Symposium, 2007. HASE '07. 10th IEEE
Conference_Location :
Plano, TX
Print_ISBN :
978-0-7695-3043-7
DOI :
10.1109/HASE.2007.42