Analysis And Defense Of Malicious Behavior For Android Software

In recent years,the use of mobile devices has become increasingly frequent and people are faced with a large number of mobile applications in their daily lives.Some data shows that more than 85% of online smart devices are now equipped with Android.As Android itself is open source,applications(Apps)on Android become the preferred target when attackers choose to attack.In addition to this,the quality of apps on the app market varies and many apps apply certain sensitive permissions during the installation process.If such apps are granted these sensitive permissions,the user's private data can be stolen with impunity,including but not limited to data collected by hardware sensors,contacts,data messages,call logs and other personal data on the user's device.To address the security issues of Android applications,this paper first analyses Android malicious applications,summarize the common Android malicious behaviors of these applications,and based on this,examines the classification and detection of Android malicious applications.The main works in this paper are:(1)a deep belief network based Android malicious application detection method is proposed;(2)a combination of static and dynamic feature extraction method is used in the feature extraction stage,and the dimensionality of extracted features is reduced by linear regression method.(3)analyzed the feasibility of applying deep learning models to Android malware detection,and constructed and implemented a classification model based on deep belief network(DBN);(4)proposed two improvement schemes based on deep confidence networks,and used sample optimization solved the problem of high complexity and long training time of deep learning models.(5)A corresponding defense strategy is proposed for Android malicious behaviors,which can be used to take corresponding preventive measures against malicious behaviors in daily life.(6)Based on the research,an Android malware detection system is designed and implemented,which allows users to upload samples to the system for detection and generates corresponding detection reports.In today's Internet era,information security issues are receiving more and more attention.This study can make people more aware of the behaviors of malicious attackers,so that they can strengthen their own security awareness and protect their privacy from being violated.