Please use this identifier to cite or link to this item:
Title: Regression Test Case Generation Technique Using Mobile Agents
Authors: Arora, Pardeep Kumar
Supervisor: Bhatia, Rajesh
Keywords: Regression test case Generation;Mobile Agents;Software Testing
Issue Date: 29-Oct-2018
Abstract: Regression testing involves testing modified software to expose faults if any introduced by the updates. It ensures that the functionality of previous code is not affected by the updates in the modified code. The focus of regression test case generation is to generate test cases for changed functionality and is a very time consuming and labor intensive task in regression testing. Most of the regression testing systems are using localized environment for test case generation, but that alone seems to be inadequate in today’s complex and rapidly changing scenario. The potential alternative to address the issues of agility and complexity seems to be distributed systems. Agent-based technology seems to be an alternative for testing multifaceted, large, complex and heterogeneous distributed systems. Agent is a portion of code that perform on behalf of a user or other program. It possess additional properties like autonomy, intelligence, proactive mobility etc. and interacts in multi-agent system. In our work, we conducted systematic literature review of existing test case generation approaches for regression testing and agent-based software testing systems. The emphasis is articulated on agent-based regression test case generation. Based on our inclusion and exclusion criteria, we identified 123 potential research papers on test case generation in regression testing and agent-based software testing and categorized them under model based testing, formal specifications based testing, structural testing, functional testing, mutation testing and random testing. The data extracted from our study are classified into seven broader areas of agent-based software testing. Based on our systematic literature survey, we recognized available techniques, approaches, platforms as well as methodologies for regression test case generation and developing agent-based software testing systems. In the first approach, we used multi-agent systems for regression test case generation on distributed environment using standard unified modelling language (UML) models and formal specifications. Different agents are designed to perform model comparison, behavior comparison, specifications comparison, impact analysis, and regression test case generation. Agents designed in Java Agent Development Environment (JADE) framework perform these tasks by using XML files of UML class diagram, sequence diagram and formal specifications xii based on Object-Z and OCL. In proposed method, all the three comparison agents: model comparison, behavior comparison, specification comparison work concurrently to save time as compared to other existing techniques. The second approach is based on the comparison of UML class diagram, use cases, activity diagram and formal specifications OCL to identify changes at both syntax and semantics level. We compared UML class diagrams, use cases and activity diagrams of old and modified code to identify these changes. It is found that the use of UML class diagram, use cases and activity diagram results in better identification of changes and hence results in efficient test case generation. Additionally, agents developed in JADE are used to collect these changes from different stake holders in the distributed environment. The distribution of testing tasks among mobile agents reduces the average time required for generation of test cases in regression testing. The third approach is based on agent based machine learning approach for identifying changes in the code for regression test case generation. These agents use object oriented metrics to identify changes within the software. In a first of its kind study, we integrated JADE with WEKA tool to design learning agents. We observe that Random Forest and Random Tree are classifiers giving best results for change impact analysis.
Appears in Collections:Doctoral Theses@CSED

Files in This Item:
File Description SizeFormat 
951003009_final.pdf3.72 MBAdobe PDFView/Open

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.