Optimization Problems for a Modular Software with Faults of Different Severity

Abstract

For the past several decades, reliability has been presumed as the most important characteristic of any complex software system. Software developer’s assurance of high quality and reliable software is driven on the basis of efficient reliability assessment. Since 1970’s an unprecedented growth has been observed in the area of software reliability growth modeling. Software reliability growth models (SRGM) are linked to the testing stage of software development and provides for insights into ways to improve reliability of the system and optimal time to release the software. Several SRGMs have been proposed in the literature to model fault identification/removal phenomenon based on time; search for more efficient and accurate models that can fit greater number of reliability growth curves is endless. Categorization of faults lying in the software has been widely studied as well. Also, efforts have been made in the past to understand reliability issues concerned with modular software system. In this paper we examine three different types of faults lying in complex software and study their behavior in an N-module software system. In order to attain two fold objectives of maximizing reliability of such a system and minimizing overall debugging cost, we propose two related optimization models. The inbuilt model flexibility takes care of different environment. The model is validated by real life software failure data set to show its goodness of fit and applicability.