Electronic Thesis/Dissertation


Prioritization of Component to Component Links for Testing in Component Based Systems Open Access

Downloadable Content

Download PDF

In recent years, with budgetary constraints and ever-changing mission requirements, the government and private industries have shifted their focus towards component based system development. The development of systems with commercial off-the-shelf components expedites the development process and improves the interoperability and reusability benefits, resulting in significant cost savings. However, integrating the commercial off-the-shelf products and testing of the integrated system is challenging. Correctly identifying dependencies between components and identifying which dependencies to test is particularly difficult for large systems. Testing every dependency is cost and time prohibitive, and budget limitations often lead to insufficient resource allocation for testing, particularly for integration testing. The key to cost-effective integration testing is identifying the type and significance of component dependencies. This dissertation proposes the use of the multi-dimensional dependency analysis to identify system components’ critical dependency, their interrelationships, and their priority based on their computed dependency strengths.This dissertation discusses the multi-dimensional dependency (MDD) analysis, a method proposed by Kafle, Sarkani, and Mazzuchi (2015) for prioritization of component-to-component links for testing based on interdependency analysis between the links. A structure matrix combined with network analysis and organizational communication analysis is used to solve complex interdependency problems without requiring previous knowledge of the dependency strength. The methodology is demonstrated through a simple application to a component based system involving a series of interdependency coefficients’ calculations to predict the critical dependency during integration and testing activities. The methodology is then validated through Markov chain analysis. Testing links are prioritized based on their interdependencies’ strengths and this is demonstrated through three case studies. The results show that the MDD method correctly identifies link priorities and is easier to use than the weighted model.These results motivate further research in the area of interdependency analysis on component links using network analysis and organizational communication analysis for the prioritization of component-to-component links in large systems.

Author Language Keyword Date created Type of Work Rights statement GW Unit Degree Advisor Committee Member(s) Persistent URL