Optimal design of fault-tolerant software systems using N-version approach and pseudo-Boolean optimization methods

Abstract

A wide range of fault-tolerant techniques has been proposed to increase the reliability of software systems, some of those techniques are algorithmic fault-tolerance, concurrent error-detection, recovery block and multiple computations. The implementation of these methods helps to avoid only physical nature (hardware) faults. When designing software support, design faults should be attended, because of their "sleeping" character (those faults originate due to the mistakes and oversights of humans that occur while they design software systems). Here, the author describes how N-version programming as an approach to fault-tolerant software system design, permits the solution of the stated problems successfully.