Extending Failure Modes and Effects Analysis Approach for Reliability Analysis at the Software Architecture Design Level

Abstract

Several reliability engineering approaches have been proposed to identify and recover from failures. A well-known and mature approach is the Failure Mode and Effect Analysis (FMEA) method that is usually utilized together with Fault Tree Analysis (FTA) to analyze and diagnose the causes of failures. Unfortunately, both approaches seem to have primarily focused on failures of hardware components and less on software components. Moreover, for utilizing FMEA and FTA very often an existing implementation of the system is required to perform the reliability analysis. We propose extensions to FMEA and FTA to utilize them for the reliability analysis of software at the architecture design level. We present the software architecture reliability analysis approach (SARAH) that incorporates the extended FMEA and FTA. The approach is illustrated using an industrial case for analyzing reliability of the software architecture of a Digital TV.

Extended Dependability Analysis of Information and Control Systems by FME(C)A-technique: Models, Procedures, Application

“…Parameter u f (v) is oriented to assessment of system operation quality decrease, description could be done using degradation diagrams, so-called QD-diagrams [22]. Thus, developed model allows assessing all dependability attributes according to the taxonomy scheme described in [17] and [20]. If we extend FME(C)A-table with a special column m f that describes possible risk reducing facilities on design (m fd ) and operation (m fo ) stages, than we get the following model:…”

“…[15][16]. Other example is that FME(C)A-approach jointly with FTA was extended for reliability analysis at the software architecture design level [17].…”

Dependability for high-tech systems: an industry-as-laboratory approach

“…This includes the use of code analysis to prioritize the warnings of a software inspection tool such as QA-C [2] and reliability analysis at the architectural level [18]. The stress testing approach of TASS artificially takes away shared resources, such as CPU or bus bandwidth, to simulate the occurrence of errors or the addition of an additional resource user.…”

Identifying critical architectural components with spectral analysis of fault trees

“…Fault trees can be used for estimating the reliability of the overall system based on individual component failures. Another goal is to estimate the sensitivity of system reliability with respect to reliabilities of system components [5,13,14]. This goal is achieved with so-called sensitivity analysis [15] or importance analysis [16,17] to identify critical components [18].…”

“…Representing a function as a sum of simpler functions allows for a sort of probabilistic reasoning about the various parameters of the system. In our approach, we use fault tree models as input, which are commonly used for sensitivity/importance analysis [2,13,14,16,18,21]. We apply spectral analysis on these models to identify critical components of the architecture.…”