Research Of Android Malware Detection Technology Based On Function Call Graph

Android operating system is an open operating system and loved by the majority of users. But Android operating system is so relatively simple security, compared with the number of large platform application software based on the characteristics of IOS ,that drew many malicious attacks for the Android market.Nowadays Android operating system is facing serious security challenges, all kinds of Android application security incident has seriously affected the development of the application of Android.So how to intelligently and effectively identify the malicious software has become the focus in the field of Android security.The operation mechanism of Android system and the Android malware call graph spectrum features were analysed in this paper .And the detection technology of extraction algorithm and machine learning algorithm were proposed, which improved the efficiency of Android malware detection. The main research contents are as follows:(1) The related Android malware detection technology at home and abroad were summarized. The attack principle and detection technology of Android malware were analysed .And the function call graph feature extraction technology based on specral theory was analysed in deep.(2) An extraction algorithm based on the inherent spectrum characteristics of Android malware was researched. A model of Android malware detection based on the inherent spectrum characteristics of call graph was proposed .And the application of machine learning model was achieved automatic ,which improve the efficiency of Android malware detection technology.(3) Android malware detection model based on function call graph was designed and implememted. The overall architechture of the system was designed and studied. Feature extraction module and malicious classification detection module were analysed and designed .And the implementation scheme was gived in detail.(4) Finally,function and performance of system is eveluated. The experiments show that the system has a better performance than other Android malware detection tools.This paper mainly studied the technique of function call graph and comebines with machine learning to realize Android malware detection technology based on function call graph . In the end of the paper we evaluated the system. The experiments showed that the system was feasible and effective.