Path testing based reliability analysis framework of component based software system

Abstract

With the increasing needs of the people in this generation, a large number of highly featured quality software systems need to be developed in all the domains. Nowadays complexity of the software system is gradually increased because of its large size. With this nature, the traditional software development process unable to produce higher quality software system within limited resources. So that, the traditional software development process has been moved to the reuse based component based software development (CBSD) which reduces the time and resource of software development. Testing is the important process in the software development life cycle to ensure the reliability or quality of software systems. Lots of reliability models have been developed to predict the software system reliability in the earlier stages of development. But these existing reliability analysis models are insufficient to estimate the reliability of component based software system (CBSS) within the limited resources. To solve this issue, the new approach was introduced by many researchers based on software architecture to estimate the reliability of component based software system. Based on that, we have proposed new framework centered on path testing to predict the reliability of the CBSS. Here we have chosen three test paths (simple, medium and complex structure) from the system for reliability estimation instead of taking all the paths. Then independent simple paths have been identified from the chosen medium and complex path to reduce the complexity of reliability estimation. All the simple paths are executed sequentially to estimate its reliability. Actual software system reliability will be predicted based on the estimated path reliability. The ATM case study has been taken to validate the proposed framework. The result obtained from this experiment is compared with the standard baseline models CUORM, LCBRM and Chao-Jung to prove the accuracy and efficiency of our proposed model. The result shows that, our proposed framework has the acceptable accuracy compared to the other models.