Research And Implementation Of Android Application Third Party Library Detection System

The third-party library of Android applications often contains hidden vulnerabilities while providing supporting applications and solutions,which brings security risks to developers and users.The existing research based on white list and clustering can not identify the third-party library with less usage,and can not effectively combat the confusion of identifier class.Therefore,this thesis proposes a detection scheme based on core logic class,which includes two parts:core logic class extraction algorithm based on PageRank and multi round matching algorithm based on sliding window The main research results are as follows:1)For the existing research,the detection efficiency of the third-party library renaming confusion is low.A detection scheme based on the core logic class is proposed to enhance the detection efficiency.The core logic class extraction algorithm based on PageRank is to transform the analysis of the whole library into the analysis of a single specific class in the library,so as to improve the detection speed.Firstly,it extracts the list of all the third-party libraries according to the Android program structure,and selects the renamed obfuscation library;then it extracts the core logic class from the obfuscation library by using the core logic class extraction algorithm,and summarizes the Dalvik instruction set features in the core logic class.2)For the problem of low accuracy in the detection of renaming confusion in the third-party library,a multi round matching algorithm based on sliding window is proposed to enhance the accuracy of detection.The matching method refers to the use of sliding window idea,through round by round comparison,to achieve accurate matching.By comparing the instruction set characteristics of the renamed obfuscated library with those of the recognizable library round by round,the sliding window is used when comparing the rows.Finally,the known library with the highest matching degree is found,and the original name of the obfuscated library is restored.Through experiments with different numbers of confusion samples,it is found that the detection accuracy of the proposed scheme is stable at 90%,and the average detection time of a single third-party library is 1.3s-1.5s,which has high accuracy and detection efficiency.3)Based on the practical problems of the third-party library,such as the rapid update rate and the miscellaneous market of the third-party library,and considering the market demand,a third-party library detection system is designed,which includes five parts:user interaction module,task scheduling module,third-party library collection module,third-party library detection module and persistent storage module.The system uses the professional level Vue and spring framework to build the front and back framework respectively,and designs the redis message queue for task scheduling,which makes the whole system organically unified.At the same time,based on this design,the third-party library detection system based on the core logic class is implemented.The system provides services for users through web pages,and can use the system to achieve fast and accurate third-party library detection.The system uses quartz cluster and redis database for task scheduling,provides users with third-party library detection function through the system,and displays detection information and detection results through a variety of data and multiple angles.