Research And Design Of Multimodal Security Detection System Based On Android Platform

In recent years, with the rapid development of Mobile Internet and the decline ofsmartphone production cost, many computer functions are slowly migrating to themobile terminal,which makes smartphones more powerful.Compared to the computer,the smart terminal is closer to users and becomes an integral part of people’s life andwork.Smartphones take advantage of portability to meet people’s needs of gettinginformation and socializing at any time.However,as the smartphone penetrationincreases, they also become attack targets of hackers and malicious software. Amongall the smartphone platforms, Android platform is suffering the most serious threatfrom malwares.In this paper,we take the Android platform as the research object, analyse theAndroid operating system’s security mechanism, and summarize the Android platformsecurity scheme and the corresponding malware detection methods at present.In onehand,through the analysis of the characteristics of mobile phones and computers,weapply the malware detection method in the field of computer to the Android platformand design a malware detection method on Android permission in full considerationof characteristics of the Android platform;On another hand,we design a multimodalsecurity system based on Android platform by digging the application requirement ofdifferent users for security system,and fully expanding usage scenarios of securitysystem.In this paper, the main work includes:(1) This paper designs a multimodal security system based on Androidplatform.This system can meet different users’ needs, and can switch to differentdetection modes flexibly under different scenarios.This System contains terminaldetection module, local PC detection module, and the cloud detection module, userof this system can achieve effective detection on his phone no matter what kind of testmode he chooses.(2) A local PC detection method based on the ADB (Android Debug Bridge) isproposed in this paper, it can transfer the APK file from mobile phone to personalcomputer and complete the process of detection on personal computer when usershave no detection software installed on the phone and can’t get access to the network.This system overcomes the default that when the mobile phone can’t connect to theInternet, the traditional security system based on the cloud detection will not be able to use.(3) Through the use of relevant decompilers, this paper decompiles the APK, andget the static analysis report.(4) Through the study of Android permission control mechanism,this paperdesigns a kind of malware detection method using machine learning method based onAndroid permission feature.Due to the limited resources of smartphones, thisalgorithm uses Principal Component Analysis algorithm for dimensionality reduction,and the detection accuracy of unknown samples is92.5%, the rate of false positives is7.5%; PC does not have this limition,and detection accuracy of unknown samples is94.05%, the rate of false positives is6%without dimensionality reduction.