Research On Memory Leak Detection Methods Based On Android

According to App Figures statistics,in 2014,the number of the applications and developers in Google Play Store exceeded the Apple App Store for the first time.The number of applications in Google Play Store has reached 1 million 430 thousand.In recent years,the market share of Android continues to rise.However,many Android applications suffer from memory leak during the practical application process.The application memory leak is a gradual process of accumulation,if not be found in time will lead to serious consequences,for example,the application OOM crash which will bring huge losses to users.Aiming at the current research status of memory leak detection methods in Android applications,combined with two main memory leak mechanisms in Android programs:Activity defects;Not closed resource objects defects,this thesis presents an Android memory leak detection model: Q-R model.The "Q" model can identify the Activity defect code by analyzing the reference types in the program,using memory growth model testing,the "R" model can identify memory defect which caused by resource object not be closed.Q-R model mainly uses three key technologies: keywords matching;dependency recognition;directional recognition.These three key technologies can be used to solve the problems of keyword feature matching,dependency orientation analysis,context relation analysis and so on.Then,using Q-R model,this paper designs a SneakFinder vulnerability detection system based on Q-R model,SneakFinder vulnerability detection system has three core modules: GapMonitor,monitoring module for monitoring memory leak;DetailAnalyzer,analysis module,for analyzing the leak type and the detailed information about the leak;Feedback,feedback module,for taking the test results back to the user.Finally,using SneakFinder vulnerability detection system combines the theory and practice of Q-R mode,perform the experiment at the PC terminal and Android mobile terminal to test whether SneakFinder can reporting the memory leak problem in system timely and effective,take the correct rate,false positive rate and false negative rate as the evaluation criteria,compare the Android vulnerability detection system SneakFinder based on Q-R model with the Android memory analysis tool MAT.Experimental results show that SneakFinder vulnerability detection system based on Q-R model can effectively detect the memory leak of Android applications,it can not only detect the known memory leak of applications,but also can detect memory leak that has not yet been found by testeddevelopers,and can effectively reduce the false positive rate and false negative rate,also improve the detection efficiency.Experiments show that the SneakFinder vulnerability detection system based on Q-R model can provide great help and support in the Android application development process.