Android Application Runtime Memory Performance Analysis Framework

Abstract

Application developers are facing various challenges during the development of applications for smartphone mobile devices. Smartphone devices have limited memory resources that restrict developers from developing computation-intensive applications. To fulfill the needs of end users, developers must take these memory limitations into account; they should try for their applications memory performance optimization.This study provides a source code level optimization of memory utilization and automated means of testing the memory for the Android application development. Sub-areas are empirically examined in detail: the impacts of resource leak patterns (AsyncTask, Thread, Handler, Static Variable, Static Variable to an Instance of Inner Class, Singleton, and Static View Activity) on memory, and the effects of these patterns on the total memory usage of an app. An empirical evaluation was conducted to see the effects of these patterns on emulators and real-time mobile devices with the two core runtime libraries (Dalvik and ART).The result shows that memory resource leakage can be avoided using the best practices suggested in the Android framework. Leak pattern can impose huge memory consumption when it runs with other activities and processes in the application. The leakage through long-running activity on the running application measures the average of 6.02 MB and 4.71 MB which is larger than the leak resulting from static reference (4.37 MB and 2.51 MB) in the emulator and real devices respectively. Moreover, the leakage due to long-running activity leak pattern and static view activity leak pattern lasts throughout the lifetime of the application.