SOFTWARE QUALITY ASSURANCE USING ADAPTIVE AGENT-BASED CLEANROOM APPROACH

Abstract

Cleanroom software engineering is a theory-based, team-oriented process for development and certification of high-reliability software systems under statistical quality control. Cleanroom combines mathematically-based methods of software specification, design. The principal objective of the cleanroom process is to develop software that exhibits zero failures in use. In this research, a study of software quality assurance and cleanroom software engineering was carried out. LoadRunner, a well-known performance and load testing software for examining software behaviour was adopted to simulate virtual users for software quality and to collect the results returned from the server side during the software quality measurement.International Business Machines (IBM)’s open source Aglet technology was also adopted to build mobile agents. All Aglets were hosted in a server which provides an environment for Aglet to execute and Graphical Users Interface (GUI) for users to control all aglets that are either created locally or migrate from a remote host. Both LoadRunner and Aglet were integrated to build a software quality measurement platform. In this model, the adaptive agent has three states and which represent Start, Usage and Final states respectively. The adaptive agent when migrates over any application software system uses eleven factors to determine the quality of the application software. Aglet was also created at a local host carrying software quality metrics. The Adaptive agent when migrate from state to state pave way for the application software testing. The factors of software quality to be considered when testing are where () correctness, () maintainability,() reliability, () usability, () testability, () flexibility, among others. Aglets Software Development Kit (ASDK), which is an open source mobile agent development package, mint Linux operating system, Ubuntu Linux operating system and window 7 operating system were the tool for the implementation. During the implementation, five standard Algorithms were used RSA, ECC, LCS, KNAPSACK and BINARY SEARCH. In order to get effectiveness of the agent operation, another method of software testing known as procedural testing was also considered. The result of this model revealed that agent method is more efficient, with reduction in time usage and less memory utilization compared to the procedural method of software testing. Because the system also involve multiple agents, fault tolerance also is enhanced because idle agents can take up the challenges in case there is breakdown of any agent during testing.