Detection And Implementation Of Android Malware Based On CNN-GRU

In recent years,with the rapid development of Internet of Things technology,the use of PC terminals and mobile terminals is also increasing exponentially.Portable mobile terminals are becoming more and more popular among people,which has brought great changes and influences on People's Daily life.In the recent years of Android smartphone promotion,QQ,wechat,email and other communication software has also been deep into People's Daily life,used for timely communication,mutual friendship;Nowadays,with the promotion of electronic payment and the health code policy implemented during the COVID-19 outbreak,everyone has become inseparable from smart phones.Different from the traditional PC devices we used before,compared with the traditional PC devices,which are bulky and difficult to carry,the smart terminal products are small in size and easy to carry.It is convenient for people to carry out social communication and other civilized activities anytime and anywhere.Compared with traditional PC terminal devices,People's Daily learning and life are more closely connected with mobile intelligent terminal devices,which not only have important information of electronic ID card,but also each user's chat information and password and other privacy things.Because of the profitdriven relationship,more and more malicious software manufacturers and hackers produce a large number of malicious software and viruses,from which they gain huge profits.Every year,a large number of users are poisoned by malicious software.Therefore,from the perspective of Android software program security,the research and improvement of malware detection technology is of great significance.In the face of more and more malicious software threats,researchers put forward corresponding solutions,based on static characteristics and dynamic characteristics of malicious software detection technology emerged.In order to improve the detection efficiency of Android malicious software,a malicious code detection scheme based on CNN-GRU model was proposed by utilizing the ability of one-dimensional CNN to deal with timing problems quickly and GRU model to deal with the long-term dependency of context.The main research contents and achievements of this paper are as follows:(1)Research and analysis of the existing malicious Android software detection technology and the architecture of the Android system.This paper summarizes and combs the existing research on Android malware detection,and conducts a comprehensive and in-depth study on the characteristics of Android system architecture,such as hardware components,App components,request permissions,API calls and so on.(2)An Android malware detection model based on CNN-GRU is proposed.Feature extraction of Android malicious code is to extract API calls,permissions,components and other relevant information from the decompilated APK file to form the final input features,and build cn N-GRU network model for detection.(3)Verify the effect of the CNN-GRU based Android malware detection model proposed in the paper,and conduct comparative experiments with other models.The detection results show that the proposed network model is efficient and accurate.By extracting static features from APK files and using CNN-GRU deep learning network model,this paper distinguishes Android software from DREBIN data set,and proposes a scheme for Android malware detection.