Assessing Diagnostic Techniques for Fault Tolerance in Software

Abstract

One of the main concerns in software safety critical applications is to ensure sufficient reliability if one cannot prove the absence of faults. Fault tolerance (FT) provides a plausible method for improving reliability claims in the presence of systematic failures in software. It is plausible that some software FT techniques offer increased protection than others. However, the extent of claims that can be made for different FT software architectures remains unclear. We investigate an approach to FT that integrates data diversity (DD) assertions and traditional assertions (TA). We also present the principles of a method to assess the effectiveness of the approach. The aim of this approach is to make it possible to evolve more powerful FT and thereby improve reliability. This is a step towards the aim of understanding the effectiveness of FT safety-critical applications and thus making it easier to use FT in safety arguments.