| Popis: |
Mobile applications (apps) are growing tremendously in the market. Many apps are useful for our day-to-day activities. In order to survive in the market, these apps must be highly responsive, and free of bugs. For an app to be popular among users, developers must focus on its responsiveness. If the response time of an app is more than the human perceivable delay (100 ms), then the user will feel the app to be slow or unresponsive. This is called the `hang bug’ problem. Developers must ensure that apps are programmed well so as to detect and correct hang bugs. Most of the existing work on hang bugs in mobile apps uses the offline detection technique i.e., detecting bugs before an app is released in the market. Hence, some bugs might exist at runtime when the user interacts with the app. If the offline technique is used, the hang bug problem will remain unknown to the developers until after release, when it will be too late to change. In this thesis, an efficient algorithm has been proposed to detect hang bugs in mobile apps during runtime, as opposed to the offline technique. The algorithm efficiently detects the API/method that is the root cause of the hang bug. There are six algorithms proposed to efficiently detect hang bugs with least delay, memory and CPU utilization. The various cases were tested on real world apps, and results were compared to find the most efficient of the cases for implementation. |
