Research On Security Analysis Techniques Of Android Applications

For today's society,intelligent terminal as the representative of the mobile Internet devices,are gradually replacing the traditional personal computers and becoming a convenient way for people to obtain information.Since Android terminals occupy a large scale of market share,more and more unlawful hackers are developing malicious applications for profit.Accurate and efficient analysis of the Android applications' security is significantly important for the development of information security research technology,and especially for the Chinese Android system users who can only visit the third party application store.This dissertation researches Android application security analysis technology.We summarize and contrast the mainstream Android application security analysis techniques and we find that the advantages and disadvantages of static analysis and dynamic analysis are complementary.This paper proposed an Android application security analysis technology thinking combine of static analysis and dynamic.In the stage of static analysis,this dissertation presents a static analysis method based on D-S evidence theory.The method can characterize application characteristics of information form characteristic index,and weight using AHP to calculate each index with malicious properties for the object of Android malware,finally calculating the Belief Function.This method has a better ability to deal with the uncertainty caused by randomness and fuzziness.In the stage of dynamic analysis,it proposes an improved KNN classification method,which is based on the cosine similarity to generate the distance.It uses time to correct the distance.Combined with the sandbox detection environment and event input technology,the method collects the characteristic runtime data,and then integrated data from the stage of static analysis to form multi attributes feature vectors.Finally we use the improved KNN classifier for classification.After an analysis process with 200 normal app and 404 malware samples,the classifier identifies malware with an accuracy of 87.5%.The simulation results show that the proposed method has a outperformed analysis effect.Based on the proposed method,this dissertation designs an Android application security analysis system(AASA).The system consists of three modules: the repackaging analysis module,the static analysis module and the dynamic analysis module.Finally,a prototype system is implemented,which can upload the APK files on the client and analyses the security of the application.At the meantime,the feedback are also returned to users.