A Fault-counting software reliability model based on a continuous-time Markov chain /
Current software reliability growth models assume that failures in software are due primarily to design faults. Hence, software failures occur most often when a program is exposed to an operational environment for which it was not intended. Though we recognize that which it was not intended. Though...
| Main Author: | |
|---|---|
| Format: | Thesis Book |
| Language: | English |
| Published: |
[Place of publication not identified] :
[publisher not identified] ;
1998.
|
| Subjects: | |
| Online Access: | http://proxy.library.tamu.edu/login?url=http://proquest.umi.com/pqdweb?did=732838261&sid=1&Fmt=2&clientId=2945&RQT=309&VName=PQD |
| Summary: | Current software reliability growth models assume that failures in software are due primarily to design faults. Hence, software failures occur most often when a program is exposed to an operational environment for which it was not intended. Though we recognize that which it was not intended. Though we recognize that software lifecycle phases of requirement and design, we wait until after the product is coded before we ''and'' the software faults. Currently, the means to approximate the resident number of faults in a product before the testing phase is derived with the use of static metrics which evaluates physical characteristics of the impending software product. An alternative to the use of static metrics is proposed and a dynamic reliability model which counts the number of faults during the software development lifecycle is presented. A single framework which exploits Markov analysis for predicting the remaining number of faults in a software product at intermediate steps throughout the entire software development lifecycle is developed. The fault counting model provides future distributions of product faults by type, i.e. requirement faults, design faults, or coding faults, at selected time intervals throughout the lifecycle prior to release of the software product to the customer. The benefits of the success of such a model are numerous. Specifically, the resulting matrices of the Markov process are immediate sources for all of the reliability measures of interest. The rapid availability of measurements which accurately reflect the changing production environment allows project managers to quickly and competently reallocate resources where most needed. |
|---|---|
| Item Description: | Vita. |
| Physical Description: | xiv, 244 leaves : illustrations ; 28 cm. Issued also on microfiche from University Microfilm Inc. |
| Bibliography: | Includes bibliographical references: pages 144-161. |