Research On Power Consumption For Android Mobile Applications Based On Program Dynamic Analysis

With the advent of the era of mobile internet, mobile communications equipmentshave become an indispensable part of our lives. In recent years, the function ofsmartphone is more and more plentiful. Besides the original calls, it has equippedwith the communication and data processing platform, including wireless internet,game applications, high resolution camera, mobile video and so on. However, withthe increase of mobile phone applications, the power consumption of mobile phone isbecoming a major obstacle restricting its development. Under the case that physicalstorage capacity of the battery is becoming to reach its limit, how to save power fromsoftware is an urgent problem that needs to solve.This paper starts from function level, and focuses on the relation betweenfunction and power consumption. It will find out the function of the biggest powerconsumption, where will most probably be energy bug. In order to do so, by differentways based on different versions of the Android system, this article designs anddevelops a detector for testing power consumption. And this work does a largenumber of experiments by the detector; tests the power consumption of the play, videoand so on; obtains the number of each function used in program in a unit test, andobtain the values of the power cost consumed by experimental testing. Then establisha test matrix as the training set. Based on the statistical experimental data, by usingSPSS (Statistics Package for Social Science) and Excel spreadsheet programs, thispaper will do some work, including the variable selection, regression equation andsignificance test for the test matrix. And a multiple linear regression model isestablished. Finally, this work finds out the function of the most power consuming byusing multiple linear regression models.The analysis result of test statistic shows that the linear regression models aresatisfactory in significance level, which includes F-test, t-test and R-test. Its efficiencyand feasibility have been proved according the three tests. This result can provide theoptimization gist, and this method proves a new idea for reducing the power consumption.