Date of Award

2014

Document Type

Thesis

Degree Name

Master of Science

School

College of Arts and Sciences

Program

Engineering & Computer Science MS

First Advisor

Roy Villfane

Abstract

Problem. The composite model approach that follows a DTMC process with constant failure rate is not analytically tractable for improving its method of solution for estimating software reliability. In this case, a hierarchical approach is preferred to improve accuracy for the method of solution for estimating reliability. Very few studies have been conducted on heterogeneous architecture-based software reliability, and those that have been done use the composite model for reliability estimation. To my knowledge, no research has been done where a hierarchical approach is taken to estimate heterogeneous architecture-based software reliability. This paper explores the use and effectiveness of a hierarchical framework to estimate heterogeneous architecture-based software reliability. -- Method. Concepts of reliability and reliability prediction models for heterogeneous software architecture were surveyed. The different architectural styles were identified as batch-sequential, parallel filter, fault tolerance, and call and return. A method for evaluating these four styles solely on the basis of transition probability was proposed. Four case studies were selected from similar researches which have been done to test the effectiveness of the proposed hierarchical framework. The study assumes that the method of extracting the information about the software architecture was accurate and that the actual reliability of the systems used were free of software errors. -- Results. The percentage difference in results of the reliability estimated by the proposed hierarchical framework compared with the actual reliability was 5.12%, 11.09%, 0.82%, and 52.14% for Cases 1, 2, 3, and 4 respectively. The proposed hierarchical framework did not work for Case 4, which showed much higher values in component utilization and therefore higher interactions between components when compared with the other cases. -- Conclusions. The proposed hierarchical framework generally showed close comparison with the actual reliability of the software systems used in the case studies. However, the results obtained by the proposed hierarchical framework compared to the actual reliability were in disagreement for Case 4. This is due to the higher component interactions in Case 4 when compared with other cases and showed that there are limitations to the extent to which the proposed hierarchical framework can be applied. The reasoning for the limitations of the hierarchical approach has not been cited in any research on the subject matter. Even with the limitations, the hierarchical framework for estimating heterogeneous architecture-based software reliability can still be applied when high accuracy is not required and not too high interactions among components in the software system exist. Thesis (M.S.) -- Andrews University, College of Arts and Sciences, 2014

Subject Area

Software architecture., Computers--Reliability.

Share

COinS