Research On Mobile APP Security Technology

The continuous improvement of mobile communication technology promoted the rapid development of mobile Internet,powerful mobile smart devices are contacted closely with people's daily lives and offer various humanistic services.Android operating system developed rapidly and occupied most market share with the open source of the system and the appeal of the market,and bred a large number of malicious applications due to its open source system.The security of mobile communication applications is directly related to the privacy and property security of users.Once users download and install malicious apps on their smart phones,malicious apps can cause privacy leaks and extra cost.Therefore,how to improve the malicious application detection performance based on Android system and how to extract more effective malware features from the application are the main subjects in the research of mobile communication APP security technology.At present,malware application detection for android mobile application which based on machine learning focuses on single feature.It could not make the best of multi-class features in Android application and could not select the optimal algorithm of different machine learning algorithms for the behavior features.This paper use D-S evidence theory to fuse the prediction results innovatively and develop mobile APP detection tools on Android platform.Firstly,this paper introduced the research background and significance of Android malware detection technology,and expounded the Android system and two main malware application detection technologies.Secondly,we extracted three different classes of Android features by calling functions of Androguard framework in python,which could reflect malicious behaviors,such as combination of permissions,system API calls and related components.Thirdly,with these three classes of static features,we chose optimal classifiers in three machine learning algorithms and fused the results of optimal classifiers to identify the Android malicious apps by D-S evidence theory.Fourthly,we designed the multi-class features fusion model and designed experiments to detect the samples of 3,657 applications which showed that the performance of this model was better.Finally,we developed the pilot health management system with the multi-feature model and the APP detection function realized the automatic detection of mobile APP.