Research On Energy Bug Judgment For Android Applications

Smartphones developed prosperly in the past few decades,but the progress of battery technology is very slow.Improving the battery life and optimizing the energy consumption of smart phones are increasingly significant and have been a hot topic of research all the time.According to statistics in recent years,the energy consumption problem of the Android system is particularly prominent.Therefore,research on energy consumption for Android applications is the key to improving the energy consumption of the Android system.Android system uses a aggressive Wakelock power management mechanism,and improper usage of Wakelock makes it easy for the system to go into a sleepless state,causing a No-sleep bug.According to research statistics,No-sleep Bug is the main cause of power loss.Therefore,the determination of the vulnerability is very important to optimize the energy consumption level.This article first obtained more than 5300 Android application software,and used the energy-saving information collection prototype MonkeyEnergyTest under stress test to collect the energy consumption data and Wakelock call information of more than 5000 application software.Secondly,statistical analysis is performed on these data and the application software with the vulnerability is selected according to the No-sleep Bug threshold.By studying the Wakelock mechanism and the No-sleep Bug energy leak,it was determined that the vulnerability caused by improper use of Wakelock was directly related to the software code.Then use the bytecode image technology to transcode the DEX file which containing the bug and no bug application to obtain the bytecode image of the corresponding application.Then the features of these images are trained using a convolutional neural network and a No-sleep Bug decision model based on bytecode images is established.Finally,using the model and the reference method to classify the test data,it is proved that the method is more versatile and convenient than the reference method,which provides a new idea for energy consumption research of application software.