Distribution Based Change-Point Problem With Two Types of Imperfect Debugging in Software Reliability

Software testing is an important phase of softwaredevelopment life cycle. It controls the quality of softwareproduct. Due to the complexity of software system andincomplete understanding of software, the testing team maynot be able to remove/correct the fault perfectly onobservation/detection of a failure and the original fault mayremain resulting in a phenomenon known as imperfectdebugging, or get replaced by another fault causing faultgeneration. In case of imperfect debugging, the fault contentof the software remains same while in case of faultgeneration, the fault content increases as the testingprogresses and removal/correction results in introduction ofnew faults while removing/correcting old ones. Duringsoftware testing fault detection /correction rate may not besame throughout the whole testing process, but it maychange at any time moment. In the literature varioussoftware reliability models have been proposedincorporating change-point concept. In this paper wepropose a distribution based change-point problem with twotypes of imperfect debugging in software reliability. Themodels developed have been validated and verified usingreal data sets. Estimated Parameters and comparisoncriteria results have also been presented