期刊名称：International Journal of Computer Science and Network Security
印刷版ISSN：1738-7906
出版年度：2018
卷号：18
期号：7
页码：38-46
出版社：International Journal of Computer Science and Network Security
摘要：Android has been effectively adopted as an open source operating system over the smart devices since it offers customers a wide range of applications. The statistics regarding number of active applications in Google Play Store show overwhelming increase. Until December 2017, the number of available applications in the Google Play Store was 3.5 million while 50.6 million number of active applications are predicted by 2020. However, there are reports of intruded applications which violates user’s privacy. It is essential to devise effective techniques to analyze and detect threats. to ensure integrity of data and applications, security experts presented various approaches including use sequences of permissions required by applications similarly system calls generated by applications are measured. This study proposes to consider intents initiated by applications as a parameter to verify malignant behavior of applications. To meet the purpose, a dataset containing 60,000 applications is generated which includes 20,000 malicious while 40,000 benign applications. The dataset is utilized to train proposed deep machine learning models including SVM and Generative Adversarial Networks (GANs). The results show reasonable malicious detection rate using intents on GANs. We believe that the proposed model is appropriate solution for ensuring security of Android applications.
