Design And Implementation Of Android Application Risk Assessment System

With the continuous development of the mobile market,the functions of mobile devices are becoming more and more abundant,and their application scenarios are also becoming more and more widespread.In all mobile device operating systems,the usage rate of Android is much larger than that of other mobile phone operating systems.According to 2017 mobile operating system sales data,Android operating system market share continued to grow year-on-year,accounting for 87.2% of the total mobile device sales share.Open source is an important reason why Android is so popular,but it is also a double-edged sword.According to related survey,99% of malware appears in Android operating system.Although the growth rate of malwares has declined in recent years,but its total amount is still very large.Therefore,research work on Android malware detection and risk assessment must be urgent.Based on previous research work on Android application risk assessment,this paper comprehensively selects multiple features,designs a layered Android application risk assessment feature system,and then an automatic data collection system for feature data was designed and implemented,which greatly reduced the labor consumption during the experiment.The Bayesian network are expanded and improved as a prototype.With basic characteristics as starting points and malicious behaviors as end points,a risk probability transfer relationship diagram is constructed,reflecting the causal relationship of risk transfer.The main research work of this thesis is as follows:(1)An Android application risk assessment system is designed,including feature extraction module and risk assessment module.With this risk assessment system,we can calculate the risk value of Android application,and user will have an intuitive understanding and reference to the risk of the Android application,thereby improving the security of users.(2)We select features commonly used in previous research work,and introduce some new features,a hierarchical Android application risk assessment feature system is established.Features included in this feature system have a great relationship with the risk of Android applications,and it is very helpful to comprehensively evaluate the risk value of Android applications.(3)An automatic data collection system for feature data is designed and implemented.We can install applications,run tests,extract data,uninstall applications,organize summary data,save to database with this system,and the entire process is automated.It can reduce the labor consumption during the experiment greatly,and contributes to the experimental data extraction work in related Android application risk assessment study.(4)Based on the Bayesian network diagram,we extend it to establish a probabilistic causal relationship between various features and malicious behaviors,forming a complete Bayesian probability risk model.The relationship model embodies the risk transfer relationship,which allows users to understand the root cause of the risk of Android applications,and has a certain effect on reducing the risk of Android applications.