Mobile Terminal Malware Detection Based On Android

Along with the fast evolution of Android system and its applications,Android users are confronting a danger from Android ecosystem.Android-based smartphones are now a perfect target for attackers for different purposes while users usually install applications from not known app stores,which provides various opportunities to download malicious applications.In our research,we perform a series of works to accomplish the detection of malware based on Android Mobile Terminal.Our main purpose was to evaluate the efficiency of static analysis for detecting Android malware application using Machine Learning and Deep Learning.In our research paper we talked about two approaches to detect Android malwareOne of these techniques used by black hats hackers consists in uploading malicious applications to app stores despite the presence of the filters which have the main objective to block malware applications on app stores.The first innovation in our research consists of detecting malicious Android applications by using meta-information which are accessible on the app store and in the Android Manifest.The technique is mainly constructed on a text mining procedure,that is used to extract important information from meta-data,that later are used to build efficient and very precise classifiers to spot Android malicious applications.Its key objective is to offer a fast and efficient way to help users to avoid being infected before installing the application to perform the analysis.The experiment result showed that the proposed method gives a higher accuracy,showing that it is able to classify malicious applications.As the number of new Android applications increases every day,the malware detection in Android mobile platforms has been a significant issue,since the detection has been made manually by analyzing the behavior and decompiled code of malware which are known to develop a malware signatures by hand.Our proposed method consists of using a static analysis of the raw opcode sequence from an APKs file,then apply a deep convolutional neural network(CNN)to detect Android malware.The results obtained by the experiments demonstrate the reliability of detecting Android malware using these approaches as it gives us a higher accuracy when using a large dataset.This method can be efficiently performed on a large number of APKs files in a short time if used in computers with high performances.