Application of Genetic Algorithm on Test Case Selection against State Coverage of Kernel Code

• Main Authors: Tzu-Hsiang Lin, 林子翔
• Other Authors: Farn Wang
• Format: Others
• Language: en_US
• Published: 2010
• Online Access: http://ndltd.ncl.edu.tw/handle/59183839783775356339

碩士 === 臺灣大學 === 電子工程學研究所 === 98 === We propose kernel code coverage analysis on model testing instead of testing on user application software, system software, and perform the testing on generating the commands in command line interfaces. In our approach, we can gain much higher coverage ratio than the used testing approach. We also can save more testing time or testing cost and save important information or error message when system crash occurs. We also propose an approach use a genetic algorithm to find a appropriate way to clustering those test cases, so that testers don’t need to run every test case; they just pick few test cases from every cluster to run, and they can evaluate the whole test case bases will gain how much coverage ratio. For the same target, there is no need to train again, we can just follow the rule we have found to clustering. Our technique provides a more efficient and more accurate way to analyze coverage ratio.