Research On Android Permission Based On Association Rules And Its Application

Android system is one of the most popular smart terminal systems.Since a large amount of privacy-sensitive data is stored in a smartphone,people pay more attention to its privacy and security issues.The Android system provides an application-oriented access control secure model,which used as permission mechanism,to protect the privacy of users.However,Android rarely introduced the relationship between permissions or the relationship between permission and API officially.Developers often overapply for permissions when developing applications,violates the principle of minimum permission and the original intention of protecting privacy.Our work proposes a research method for Android permission based on association rules.The association rules are mined for the permissions which actually used by the applications in each category in the current application market,and then,the features of the permissions usage of different categories' applications are obtained.At the same time,we use statistical analysis method to analyze the relationship between the API and the permissions of different system versions of Android,and infer that the degree of association between the permissions is more and more related in the process of version evolution.At present,there are many applications declare excessive permissions in the mobile application market.Therefore,this thesis designed and implemented a tool to detect the permissions actually used by applications.The tool can decompile the application's apk in mobile application markets,and statically scan the decompiled files,extract Android API used by the application,and finally obtain the application's permissions that the application actually need according to the relationship between the API and the permission of Android system.This tool can help users detect if the target application has excessive permissions claim.According to the permissions actually used in the application,this thesis uses permission research method based on association rules to summarize the characteristics of permissions used in application in different categories of application markets,designs and develops a rationality detection tool to detect application's category.This tool helps application publishers determine whether the application released in a reasonable category.The rationality detection tool is the usage point of the permission research method based on association rules.The tool judges whether the permissions used by the application conform to the permissions association rule of the application target category,and then,it detects whether the target category is rationable where the application released in.Experimental results show that the rationality detection tool can detect applications with unreasonable category selection.In addition,this thesis also mines the association rules of the permissions used by malicious applications.According to the association rules,we design and develop a evaluation tool to detect app store security.The tool can filter out suspicious applications which permission useage conform to the permissions association rule of malicious applications.Then,the security evaluation standard of the application store is the proportion of the detected suspicious applications,the lower the proportion,the safer the application store.According to the test of this thesis,the evaluation tool which can detect app store security,has a good ability to filter suspicious applications,and can measure relatively safe app stores from different app stores.